Contents of /trunk/ChangeLog

2008-09-27 R. Steven Rainwater <steve@ncc.com>

	* site.c: Removed deprecated AdInfo struct
	
	* site.c (site_render): Added experimental YouTube link. Place a 
	YouTube video ID inside like so: <youtube>d9gjy1OsTGQ</youtube>

	* util.c (virgule_youtube_link): Genrates the YouTube link code
	from the supplied video ID.
	
	* style.c (virgule_render_header): Cleaned up some cruft

	* INSTALL: Complete rewrite of the INSTALL file based on user feedback

2008-08-15 R. Steven Rainwater <steve@ncc.com>

	* acct_maint.c (acct_person_diary_rss_serve): Fixed typo in RSS 
	version number (removed extra trailing dot).


2008-08-04 R. Steven Rainwater <steve@ncc.com>

	* article.c (article_render_from_xml): replaced table-based article
	header HTML with cleaner div markup.


2008-06-27 R. Steven Rainwater <steve@ncc.com>

	* mod_virgule (set_virgule_db): db arg is now a const 
	
	* mod_virgule (read_site_config): fixed cases where a certain 
	config.xml setup could result in a strlen(NULL).
	
	* Makefile: Cleaned up a and added comments about Apache ARP 1.x

	* acct_maint.c: Added xmlChar/char casts to eliminate gcc 4.x warnings.
	* diary.c: Added xmlChar/char casts to eliminate gcc 4.x warnings.
	* db_ops.c:  Added xmlChar/char casts to eliminate gcc 4.x warnings.
	* db_xml.c:  Added xmlChar/char casts to eliminate gcc 4.x warnings.
	* tmetric.c: Added xmlChar/char casts to eliminate gcc 4.x warnings.
	* schema.c: Added xmlChar/char casts to eliminate gcc 4.x warnings.
	* site.c: Added xmlChar/char casts to eliminate gcc 4.x warnings.
	* util.c: Added xmlChar/char casts to eliminate gcc 4.x warnings.


2008-01-23 R. Steven Rainwater <steve@ncc.com> 

	* diary.c (virgule_diary_entry_render): Added call to 
	virgule_nice_htext before rendering diary. This is a kluge to get
	around the increasingly nasty HTML syndicated from external blogs.
	This should be replaced with proper HTML filtering and repair in the
	aggregator storage module (it will also be necessary to retroactively
	repair stored blogs before removing this code).


2008-01-09 Mauro Persano <mauro_persano@yahoo.com>

	* mod_virgule.c (read_allowed_tag): Added this function to parse
	allowed tag structures in the config.xml file.
	
	* mod_virgule.c (read_site_config): Modified to handle tag attribute
	data from config.xml.
	
	* util.c (virgule_add_allowed_tag): Modified to handle tag attribute 
	data from config.xml.
	
	* util.c (nice_tag): Added this function to filter out disallowed
	tag attributes in HTML.
	
	* util.c (nice_htext): Added call to nice_tag()


2007-12-15 R. Steven Rainwater <steve@ncc.com>

	* db.c (virgule_db_lock_key, virgule_db_lock_upgrade, 
	virgule_db_lock_downgrade, virgule_db_lock, virgule_db_unlock):
	Removed all the existing lock functions and all references to them
	throughout mod_virgule. Instead of locking the entire database, 
	these calls are replaced with Apache APR file locking calls to 
	lock only the specific DB Key (file) being read or written. This 
	makes mod_virgule much more responsive and a bit more scalable.
	
	* db.c (virgule_db_del): Replaced unlink() with apr_file_remove().
	Replaced rmdir() with apr_dir_remove().
	
	* db.c (virgule_db_put_p): Added exclusive apr_lock_file() call.
	
	* db.c (virgule_db_get_p): Added shared apr_lock_file() call.


2007-12-05 R. Steven Rainwater <steve@ncc.com>

	* diary.c (virgule_diary_rss_export): The guid of syndicated diary
	entries, when available, is preserved in Advogato's outbound diary
	RSS feed. This should help downstream aggregators identify duplicate
	posts more easily. Thanks to Paul Wise for noticing this.


2007-11-07 R. Steven Rainwater <steve@ncc.com>

	* private.h: Added pointer to a Google Analytics string

	* mod_virgule.c (read_site_config, info_page): Added support for a 
	Google Analytics code in the config.xml file
	
	* style.c (virgule_render_footer_send): Added support for Google
	Analytics. If a GA ID is found, the analytics tracking code is
	inserted prior to the HTML body close tag.

	* acct_maint.c (virgule_acct_update_art_index): Reduced memory
	leakage substantially.


2007-11-04 Joe Presbrey <presbrey@mit.edu>

	* acct_maint.c (prof_fields): Added field in profile for a FOAF URI
	
	* foaf.c (virgule_foaf_person): Added xmlns:owl namespace to FOAF. 
	An owl:sameAs tag is now added for the FOAF URI, if present.


2007-11-02 R. Steven Rainwater <steve@ncc.com>

	* config.xml: Fixed XML well-formedness error. Thanks to presbrey
	for spotting this and the template issues.
	
	* sample_db: Added several missing generic default templates needed
	for a new install. 
	

2007-10-22 R. Steven Rainwater <steve@ncc.com>

	* foaf.c (virgule_foaf_person): Added export of the computed trust
	level for each user with reference to a local RDF schema of Advogato
	trust levels. This mechanism was suggested by Sean B. Palmer and
	Dan Connolly on the W3C #swig IRC channel.


2007-08-31 R. Steven Rainwater <steve@ncc.com>

	* foaf.c (virgule_foaf_person): The mbox_sha1sum field is not
	included if the user profile has no email address. Thanks to
	Andreas Harth of deri.org for reporting the bug.

	* site.c (virgule_site_render_person_link): A rel="nofollow" is 
	added to any person link if the user is not trusted. This should
	help minimize the value of untrusted accounts created by spammers.
	Suggested by ncm.

	* acct_maint.c (virgule_acct_person_index_serve): The user's cert
	level is evaluated and rel="nofollow" added for untrusted users.

	* site.c (site_render_recent_acct): The user's cert level is now
	passed to virgule_site_render_person_link().

	* diary.c (virgule_diary_entry_render): The user's cert level is now
	passed to virgule_site_render_person_link().

	* certs.c (virgule_render_cert_level_begin): The CertLevel value is
	now returned. This prevents the need to do multiple trust metric
	lookups in some cases.
	

2007-08-26 R. Steven Rainwater <steve@ncc.com>

	* mod_virgule (info_page): Fixed pointer dereferencing bug.

	* rating.c (rating_clean): Added a maintenance function that 
	analyzes the eigen vector cache and removes local cache files of
	users that have been deleted. Ratings of nonexistant users found
	in good cache files are also removed.
	

2007-08-14 R. Steven Rainwater <steve@ncc.com>

	* acct_maintc. (acct_person_diary_static_serve): Handles a static
	diary URL by rendering a single diary entry in fancy recentlog mode
	with the addition of two links to additional diary entries for the
	user.

	* acct_maint.c (acct_person_serve): Added handlers for static diary
	URLs. 


	* diary.c (virgule_diary_entry_render): Change diary entry permalinks
	to static URLs. This may improve search engine rankings for diary
	entries and will come in handy someday when we add comments to diaries.
	

2007-08-01 R. Steven Rainwater <steve@ncc.com>

	* util.c (virgule_render_url): URL is now escaped with ap_escape_html()
	rather than ap_os_escape_path(), which was freaking out on URLs that
	contained a ? character. This should have been done as part of the
	migration from the Apache APR 1.3 -> 2.0 API. Props to redi for 
	spotting this bug.


2007-07-29 R. Steven Rainwater <steve@ncc.com>

	* acct_maint.c (acct_person_serve): Account profiles now include a 
	list of the 10 most recent articles posted by the user and link to
	a full listing of articles.

	* acct_maint.c (virgule_acct_update_art_index): Added this function
	to create and maintain an XML index in the user account directory of
	every article the user has posted.

	* acct_maint.c (acct_person_articles_serve): Renders a full list of
	all articles posted by the specified user.

	* acct_maint.c (acct_kill): The articles.xml file is removed with a
	user account is deleted.

	* article.c (article_maint): Added this function to sequentially 
	process all articles and update user profiles with a correct index
	of articles each user has posted.

	* article.c (virgule_article_serve): Added URL for the new article
	maintenance function.


2007-07-22 R. Steven Rainwater <steve@ncc.com>

	* acct_maint.c (acct_person_serve): Inbound and outbound trust certs
	less than 30 days old are marked as "new" in the user profile.


2007-07-19 R. Steven Rainwater <steve@ncc.com>

	* foaf.c: Added xmlChar/char casts to eliminate gcc 4.x warnings.
	* proj.c: Added xmlChar/char casts to eliminate gcc 4.x warnings.
	* rss_export.c: Added xmlChar/char casts to eliminate gcc 4.x warnings.
	* xmlrpc.c: Added xmlChar/char casts to eliminate gcc 4.x warnings.


2007-07-16 R. Steven Rainwater <steve@ncc.com>

	* proj.c (proj_proj_serve): 
	* article.c (article_render_from_xml): 
	* acct_maint.c (acct_person_server): Apostrophe is now escaped in the
	HTML calling the social bookmarking Javascript function. Thanks to
	LKCL for spotting the error.


2007-07-09 R. Steven Rainwater <steve@ncc.com>

	* aggregator.c (aggregator_post_feed): Check for existing GUID before
	doing an update in case the feed has retro-actively altered the item's
	post time stamp. This should prevent item dupes on feeds that mess 
	with the time stamps in bad ways, provided the GUID remains valid.

	* article.c (article_render_from_xml): Replaced hardcoded social
	bookmark links with standard share icon and call to external 
	javascript social bookmarking widget.

	* acct_maint.c (acct_person_serve): Added social bookmark link to
	profile page. Replaced "this person" to the actual user name in the
	profile certification banner.

	* proj.c (proj_proj_serve): Added social bookmark link to project
	page. Tweak HTML slightly to improve look.


2007-07-06 R. Steven Rainwater <steve@ncc.com>

	* acct_maint.c (acct_logout_serve): The no_cache and no_local_copy
	flags are now set in the Apache request_rec structure to prevent
	browsers from caching old logout results. This bug was preventing a
	few users from successfully logging out of Advogato. Props to
	Paul Wise (pabs3) for bringing the problem to my attention.


2007-06-13 R. Steven Rainwater <steve@ncc.com>

	* v.js (clrField): Started a mod_virgule Javascript library. 
	The clrField() function allows helpful comments to be placed in
	form fields that will vanish when the user focuses the field.

	* style.c (virgule_render_header): Now includes the new Javascript
	library in each page header. Removed the hardcoded frame-buster 
	Javascript from	the page headers and put it in the v.js file, 
	allowing it to be cached across page views. This reduces the page
	size slightly.

	* util.c (virgule_strip_a): Fixed a bug that allowed spammers to
	circumvent having anchor tags stripped from their notes field
	by intentionally entering malformed markup.

	* acct_maint.c (acct_index_serve): Cleaned up the account action
	list.
	

2007-06-08 R. Steven Rainwater <steve@ncc.com>

	* db_ops.c (db_relation_put_field): Fixed a bug that prevented staff
	relations from being removed when set to a relation type of None.
	Props to Gary Benson for noticing the problem.

	* proj.c (virgule_proj_set_relation): Fixed bug that prevented
	db_relation_put_field() from being called on certain staff
	relationship changes.
	

2007-05-16 R. Steven Rainwater <steve@ncc.com>

	* util.c (virgule_force_legal_css_name): Added function to force an
	arbitrary string into compliance with the CSS1 spec for use as a CSS
	class name. This will allow usernames to be inserted into class
	statements as additional identifiers. Some third party blog handling
	utilities rely on this arrangement to improve dupe handling.

	* diary.c (virgule_diary_entry_render): Wrapped each entry in a div
	and assigned two classes, one for site design and one that's a
	CSS'ified version of the user name for screen-scraping aggregator
	use. Also switched the diary entry body to a div from a blockquote
	allowing more flexibility in site layout.


2007-05-15 R. Steven Rainwater <steve@ncc.com>

	* private.h (virgule_private): added editors list pointer. This will
	allow a site to optionally limit article posts to a specific set of
	users, known as editors. Also changed article_post_by_seeds_only to
	article_post_by_editors_only. The seed list was serving double duty
	on robots.net as an editors list but it's now helpful to separate
	the functions.

	* mod_virgule.c (read_site_config): Added code to read the list of
	editors from the config.xml file on startup.
	
	* mod_virgule.c (info_page): The diagnostics page now dumps the 
	list of editors, if any, and the current setting of the editor only
	flag read from the config.xml file.
	
	* req.c (virgule_req_ok_to_post): Now queries the editors list
	if the article_post_by_editors_only flag is set.

	* article.c (article_form_serve): The article post page now uses
	a page-specific template instead of the default template to allow
	for more site-specific customization such as posting guidelines.


2007-05-03 R. Steven Rainwater <steve@ncc.com>

	* general: Moved module templates to the new template directory.

	* style.c (virgule_render_header): Removed references to vr->raw
	flag which not needed with the streamlined header code.


	* style.c (virgule_render_footer_send): Removed reference to vr->raw,
	removed unneeded insertion of a </div> tag, Removed unneeded call
	to virgule_render_sitemap().

	* style.c (virgule_render_sitemap): Removed references to the
	unneeded vr->sitemap_rendered flag. 

	* req.h (_VirguleReq): Removed unused raw and sitemap_rendered flags

	* mod_virgule.c (virgule_handler): Removed unused reference to
	vr->sitemap_rendered flag.


2007-05-03 R. Steven Rainwater <steve@ncc.com>

	* proj.c (proj_new_serve, proj_edit_serve): Output is now rendered 
	within the default template.

	* article.c (article_render_from_xml): Reddit submissions now 
	include the article title.

	* article.c (article_form_serve): Output is now rendered within
	the default template.
	
	* article.c (article_generic_submit_serve): Output is now rendered
	within the default template.

	* article.c (article_reply_form_serve): Output is now render
	within the default template.
	
	* site.c (virgule_site_render_page): Removed code related to the
	now unneeded distinction between standard and raw page headers.

	* style.c (virgule_render_header): Removed this now unused function
	along with a large block of hard-coded, site-specific header
	markup for robots.net and Advogato. It's all in the templates now.

	* style.c (virgule_render_header_raw): Renamed this function to
	virgule_render_header().
	
	
2007-04-30 R. Steven Rainwater <steve@ncc.com>

	* acct_maint.c (account_index_serve): Account info page and account
	login page are now template based. 100% of the login page markup is 
	now configurable in the template. 

	* diary.c (diary_index_serve, diary_preview_serve): Output is now 
	rendered within a template.

	* diary.c (diary_edit_serve): Output is now rendered in a template.
	Fixed several bugs dating back to the first diary editing patch, one
	of which could potentially have been the cause of Apache segfaults.


2007-04-29 R. Steven Rainwater <steve@ncc.com>

	* site.c (site_render): Template tag replacement strings with NULL
	values are now allowed.

	* style.c (virgule_render_in_template): Templates are now rendered
	even when no content is provided for them.


2007-04-25 R. Steven Rainwater <steve@ncc.com>

	* aggregator.c (aggregator_post_feed): Items with no content are
	skipped rather than creating empty diary posts.
	
	* style.c (virgule_send_error_page): Error pages are now rendered 
	using the default.xml template. All hardcoded markup has been removed.
	An additional argument was added to allow reuse of this function for
	informational messages as well as error messages. Replaced old-style
	<tt> tags with <em> tags. The old-style look can be duplicated in CSS
	if desired.

	* style.c (virgule_render_header_raw): Now produces legal HTML even
	if title is NULL.

	* site.c (virgule_site_render_page): Improved handling of the page
	title to allow for static title text in the template, title text
	provided by mod_virgule, and combinations of the two if desired.

	* tmetric.c (tmetric_index_serve): Replaced hardcoded output page
	with an info page generated by virgule_send_error_page().

	* rating.c (rating_crank, rating_crank_all): Replaced hardcoded 
	output pages with info pages genrated by virgule_send_error_page().

	* rating.c (rating_report): Replaced hardcoded output with a 
	template based output page. The default.xml template is used.

	* aggregtator.c (aggregator_getfeeds_serve): Replaced hardcoded 
	output with a template based page. Default.xml template is used.


2007-04-17 R. Steven Rainwater <steve@ncc.com>

	* aggregator.c (aggregator_index_atom_10): Fixed handling of ATOM
	feed in which the entry containers have a <summary> tag but no 
	<content> tag. Also supported now are ATOM feeds which include 
	only an <updated> tag and not a <published> tag.


2007-04-09 R. Steven Rainwater <steve@ncc.com>

	* MAKEFILE: Changed -fPIC to -fpic. This may have been the cause of
	a couple of segfaults, due to a conflict with the flag used to compile
	Apache and/or libxml2 on some distros. The -fpic flag should be
	correct on 99% of the platforms likely to be running mod_virgule.

	* acct_maint.c (virgule_acct_touch): Added missing db lock calls.
	
	* rss_export.c (rss_index_serve, rss_render_from_xml): Changed a few
	xmlNewChild calls to xmlNewTextChild to assure proper entity escapes.

	* site.c (virgule_site_send_banner_ad, virgule_site_render_banner_ad):
	Removed these functions. They have been rendered unnecessary by the
	newer general purpose include tag which allows external banner ad
	management tools to be used.

	* mod_virgule (virgule_handler): Fixed minor aliasing bug. Removed
	call to virgule_site_send_banner_ad.


2007-04-07 R. Steven Rainwater <steve@ncc.com>

	* req.h: Added a temporary buffer for pre-rendering template data.
	Also added a header buffer which allows additional page header
	contents to be inserted easily. The old header buffer kluge that
	passed strings through virgule_render_header() has been removed.

	* mod_virgule (virgule_handler): Set the new buffer pointers
	to NULL when initializing a request structure.
	
	* style.c (virgule_set_temp_buffer, virgule_set_main_buffer): Added
	functions to switch between the request buffer and the temp buffer
	more easily. A temporary buffer is used when building pages that 
	will be rendered within a preloaded template. The main buffer is
	used to render pages directly. 
	
	* style.c (virgule_render_in_template): Added a new function to handle
	rendering an arbitrary page from a template. This collects some
	duplicate code that was being used in several places and should
	make things a little easier to maintain going forward.

	* style.c (virgule_render_header_raw): If the header buffer is not
	empty, its contents are inserted into the page header at this time.

	* style.c (virgule_send_error_page): A page not found error string
	is automatically added if the error code has been set to 404. This
	eliminates the need for lots of duplicated static strings.
	
	* acct_maint.c (acct_person_diary_serve): Replaced existing code
	with the new template rendering functions.

	* article.c (article_num_serve): Replaced existing code with the 
	new template rendering functions.

	* buffer.c (virgule_buffer_size): Added function to return the 
	current size of a buffer.
	

2007-04-05 R. Steven Rainwater <steve@ncc.com>

	* acct_maint.c (acct_person_diary_rss_serve): Added memory allocation
	check on XML document dump process to prevent segfaults.

	* acct_maint.c (acct_flag_as_spam): Removed early free since the
	normal pool cleanup occurs immediately afterwards anyway.

	* rss_export.c (rss_index_serve): Added an additional error check
	to the xml rendering process to prevent allocation error segfaults.
	
	* db_xml.c (db_xml_free): Additional sanity check before optional
	xmlFreeDoc calls that occur outside of normal APR pool life cycle.
	Also removed an unused parameter from the function call throughout
	the mod_virgule code.

	* foaf.c (virgule_foaf_person): Handles accounts with no givenname
	or surname in a safer fashion.

	* tmetric.c (tmetric_run): Removed unneeded check for NULL value
	before call to db_xml_free().


2007-03-30 R. Steven Rainwater <steve@ncc.com>

	* article.c (article_render_from_xml): Added some initial social
	bookmarking support for Digg, del.icio.us, and Reddit.
	
	* sample_db: Added PNG images for Digg, del.icio.us, and Reddit.


2007-03-21 R. Steven Rainwater <steve@ncc.com>

	* mod_virgule.c (info_page): Added a lot of additional diagnostic
	info from the Apache thread private record structure. Also moved all
	the diagnostic info into a nicely formatted table that's a lot easier
	to read.


2007-03-19 R. Steven Rainwater <steve@ncc.com>

	* acct_maint.c (acct_person_diary_serve): Blog pages are now rendered
	from a template (/site/person/diary.xml) instead of hard-coded HTML.
	A sample diary.xml template was also added to the sample XML DB.
	

2007-03-15 R. Steven Rainwater <steve@ncc.com>

	* diary.c (virgule_diary_[store|update]_feed_item): These functions
	are now able to deal with feed formats that carry the blog content
	as unescaped XML trees. At least one Blogger ATOM feed uses this
	method. 

	* util.c (virgule_sha1): Return an SHA-1 hash as a string. Accepts
	input strings of arbitrary length. Based on the Apache APR SHA-1
	implementation.

	* mod_virgule.c (info_page): Added admin email and FOAF SHA-1 test.

	* foaf.c (virgule_foaf_person): Added support for foaf:mbox_sha1sum
	field (an SHA-1 hash string of the user's email address).

2007-03-08 R. Steven Rainwater <steve@ncc.com>

	* acct_maint.c (acct_newsub_serve): The password field is now a
	required field on new accounts. The password is no longer allowed to
	be the same as the username. The email field is now required. Also 
	removed some old debugging cruft from this function.
	

2007-03-07 R. Steven Rainwater <steve@ncc.com>

	* acct_maint.c (acct_person_serve): Account profile "extra junk" 
	error was replaced with a conventional 404 error. 
	
	Requests for invalid account names now return a 404 error as well.
	
	FOAF autodiscovery header and FOAF image link now show only for
	trusted users.  
	
	RSS autodiscovery header and RSS link now show only for trusted users
	who have at least one blog post.

	* acct_maint.c (acct_person_foaf_serve): FOAF files are served only
	for trusted users. FOAF requests for untrusted users now generate a
	404 error.

	* acct_maint.c (acct_person_diary_rss_serve): RSS feeds are now 
	served only for trusted users with at least one blog post. RSS 
	requests for untrusted users or trusted users who have no blog
	posts now generate a 404 error.

	* diary.c (virgule_diary_exists): Added this function to test if
	there are any posts in a diary.
	

2007-03-05 R. Steven Rainwater <steve@ncc.com>

	* article.c (article_submit_serve): Duplicate article rejection no
	longer triggers on article edits of the most recently posted article.

	* article.c (article_render_from_xml): Now displays the article
	update timestamp for edited articles.


2007-02-24 R. Steven Rainwater <steve@ncc.com>

	* mod_virgule.c (read_site_config): Added support for articledays2edit
	config value. This will determine the number number of days after
	posting that an article remains editable by the author.

	* article.c (article_render_from_xml): Added edit link to article
	subtitle if viewed by the author within edit period.


2007-02-23 R. Steven Rainwater <steve@ncc.com>

	* article.c (virgule_article_serve): Added handler for edit function.

	* article.c (article_edit_serve): Loads an existing article into the
	editor. Only works for the original author and within a configurable
	time period.

	* article.c (article_generic_submit_serve): Added kludge to support
	editing of existing articles, preservation of original article date,
	and original article key value. This code is pretty nasty but no
	worse than the rest of kludges piled on top of each other in here.
	Functions like this one will be the first against the wall when the 
	revolution comes.
	

2007-02-21 R. Steven Rainwater <steve@ncc.com>

	* acct_maint.c (acct_person_serve): Toned down the certification
	form description a bit.


2007-02-17 R. Steven Rainwater <steve@ncc.com>

	* site.c (site_render): Added userstats tag support.

	* style.c (virgule_render_userstats): Added function to render the
	new userstats.xml data as a simple table.


2007-02-16 R. Steven Rainwater <steve@ncc.com>

	* certs.c (virgule_cert_verify_outbound): Added this function to 
	verify and repair outbound certs when called by acct_maint(). 
	Checks are performed for a variety of cert errors that are known to
	have occured on mod_virgule sites due to file corrupt. Corruption 
	that cannot be repaired is reported through a return value for 
	manual attention. 

	* certs.c (virgule_cert_verify_inbound): Added this function to 
	verify and repair inbound certs when called by acct_maint().


2007-02-12 R. Steven Rainwater <steve@ncc.com>

	* acct_maint.c (acct_kill): Feedlist is only rewritten if a record
	was removed for the deleted account. Blog feed buffer (feed.xml) is
	removed during the account deletion process.
	
	* acct_maint.c (acct_maint): Added new function to sequentially
	check all user profiles for abnormalities. Unrecoverable problems
	are reported for manual repair. Recoverable problems like asymmetric
	certification are fixed automatically.


2007-02-09 R. Steven Rainwater <steve@ncc.com>

	* style.c (virgule_render_header_raw): Removed the ancient CSS hack
	used to for pre Netscape 4 browsers. The number of additional hits
	generated by the hack is no longer worth it for backwards 
	compatibility with a browser that's not used anymore.


2007-02-08 R. Steven Rainwater <steve@ncc.com>

	* style.c (virgule_render_date): Added seconds to the timestamp
	rendering style used for RSS pubDate field. Props to 
	Wade Berrier <wberrier@novell.com> for noticing this bug. 


2007-02-07 R. Steven Rainwater <steve@ncc.com>

	* mod_virgule.c: Removed some cruft and several test functions that
	have gone unused for years. 

	* mod_virgule.c (info_page): Removed reference to count(), a test
	function which is no longer used. Cleaned up and correct HTML. Added
	additional diagnostics output.


2007-02-04 R. Steven Rainwater <steve@ncc.com>

	* site.c (site_render_recent_proj): Removed unneeded class from
	recent project div wrappers. Assigning the class in the style sheet
	is more flexible and results in smaller HTML.
	

2007-02-03 R. Steven Rainwater <steve@ncc.com>

	* foaf.c (virgule_foaf_person): Changed foaf:seeAlso to the
	correct rdfs:seeAlso. Oops. (thanks to Morten Frederiksen for
	pointing out the bug).


2007-02-02 R. Steven Rainwater <steve@ncc.com>

	* foaf.c (virgule_foaf_person): More FOAF changes (thanks go to 
	Richard Cyganiak, Benjamin Nowack, Andreas Harth and Danny Ayers).
	Output should now work better with RDF applications like Tabulator 
	and Disco. Projects are now converted to document references 
	instead of foaf:Project records, which should be more consistent 
	with the latest interpretations of the FOAF 0.1 spec.

	* acct_maint.c (acct_person_serve): Normalized and corrected the
	HTML used to describe a user's projects on the profile page. Added
	a FOAF badge and link.


2007-02-01 R. Steven Rainwater <steve@ncc.com>

	* foaf.c (virgule_foaf_person): Refactored the code a bit to move all
	the FOAF-specific code into its own file. This should make things
	easier down the line if we add FOAF files for the projects. Added a
	number of additional new properties to the FOAF output and made some
	minor corrections to the syntax based on feedback from the foaf-dev
	mailing list.

	* acct_maint.c (acct_person_foaf_serve): This now just a wrapper for
	for virgule_foaf_person that render the returned XML document.


2007-01-30 R. Steven Rainwater <steve@ncc.com>

	* acct_maint.c (acct_person_foaf_serve): Added foaf:knows properties
	for each outbound cert in a user's profile. Added foaf:Project
	properties for each project associated with the user.


2007-01-29 R. Steven Rainwater <steve@ncc.com>

	* rating.c (virgule_rating_serve): Changed the diary (blog) ranking
	crank URL from /rating/crank.html to /admin/crank-diaryrating.html 
	to move it to a password protectable area where it can't be abused.

	* aggregator.c (virgule_aggregator_serve): Changed aggregator crank
	URL from /aggregator/getfeeds.html to /admin/crank-aggregator.html
	to move it to a password protectable area where it can't be abused.


2007-01-28 R. Steven Rainwater <steve@ncc.com>

	* acct_maint.c (acct_person_serve): Added FOAF auto-discovery URL
	to the account profile page head section.
	
	* acct_maint.c (acct_person_foaf_serve): Added name, nick, homepage,
	and weblog tags to FOAF file. 

	* mod_virgule (info_page): renamed from test_page() and changed the
	URL to /admin/info.html from /foo.html. This will allow the URL to be
	added to a general password protected admin area of the site. The
	info page exposes diagnostsic information that would be helpful to 
	script kiddies and crackers.

	* mod_virgule (virgule_handler): Reordered dispatching sequence to
	put administration and less used handlers last. Highly used handlers
	such as RSS, articles, and diaries are now first. This should provide
	a (very) small performance improvement.

	* tmetric.c (virgule_tmetric_serve): Changed the trust metric crank
	URL from /tmetric/ to /admin/crank-tmetric.html to move it to a
	password protectable area where it can't be abused.


2007-01-27 R. Steven Rainwater <steve@ncc.com>

	* acct_maint.c (acct_person_serve): Added FOAF URL support. 
	Deprecated diary.xml (virgule_diary_export) URL support.

	* acct_maint.c (acct_person_foaf_serve): Generates a FOAF RDF file
	from a user profile.

	* certs.c (virgule_cert_set): Certifications now include a date and
	time stamp. These will be used in the near future.

	* diary.c (virgule_diary_export): Deprecated this function. It was
	superceded by the xml-rpc interface a long time ago, it generates
	invalid XML, it acts an effective denial of service attack vector,
	and is almost unused anyway.


2007-01-19 R. Steven Rainwater <steve@ncc.com>

	* acct_maint.c (acct_person_serve): Removed hard-coded RSS icons and
	related HTML. Replaced with div that can be configured in CSS.


2007-01-17 R. Steven Rainwater <steve@ncc.com>

	* rss_export.c (rss_render_from_xml): Added RSS 2.0 required tags,
	removed RSS 0.91 DTD reference, added several options tags such as
	guid and generator.
	
	* rss_export.c (rss_index_serve): Minor tweaks to generate RSS 2.0
	feeds.

	* acct_maint.c (acct_person_diary_rss_serve): Changed RSS version
	in XML header to 2.0.
	
	* diary.c (virgule_diary_rss_export): Converted hard-coded time zone
	to use GMT instead.  Added required RSS 2.0 tags and several optional
	one include guid and geneator. Title is now export correct if it is
	present in the diary entry. If a title is not available, the date is
	used as the title. If neither title nor date is available, the channel
	title is used for the item title.


2007-01-14 R. Steven Rainwater <steve@ncc.com>

	* util.c (virgule_iso_now): Time is returned in UTC (GMT) rather than
	server local time. By using UTC times consistently in the XML data
	store, it simplifies time calculations and becomes easier to render
	user-specific time zones and formats later on.

	* util.c (virgule_time_t_to_iso): Time is returned as UTC (GMT) rather
	than server local time.

	* util.c (virgule_virgule_to_time_t): Virgule timestamps are now
	assumed to always be in UTC (GMT) rather than server local time.

	* diary.c (virgule_diary_latest_feed_entry): Diary timestamps are
	now assumed to be in UTC (GMT) rather than server local time.

	* style.c (virgule_render_date): Time is now rendered as UTC rather
	than server local time zone.

	
2007-01-06 R. Steven Rainwater <steve@ncc.com>

	* aggregator.c (aggregator_post_feed): When an entry with new date
	is found, if the entry has a unique ID, a search done against the
	existing entries for a match. If a match is found, we assume the date
	is incorrect and perform an update on the existing entry rather than
	adding as a new entry. This should solve the problem of	duplicate 
	posts with certain RSS feeds that alter entry dates retroactively.

	* diary.c (virgule_diary_entry_id_exists): Search diary entries for
	the specified entry ID and return the entry index number.

	* diary.c (virgule_diary_update_feed_item): Added entry index number
	to parameter list. This optional value allows the entry to be
	accessed directly without a search, if it is known.

	* diary.c (virgule_diary_latest_feed_entry): Extended search for a
	recent entry with a feedpost tag to 20 more recent entries. This
	should handle any reasonable case of combined local and syndicated
	usage.


2007-01-05 R. Steven Rainwater <steve@ncc.com>

	* diary.c (virgule_diary_store_entry): Completely rewrote this code
	to make the function safe to use on syndicated entries. All the
	syndication tags are now preserved when an updated entry is stored
	through the local website or through the xml-rpc interface.

	* diary.c (diary_edit_serve): Added support for syndication-related
	fields, so we can warn users to be careful editing syndicated posts.
	Also added initial support for the title field.


2007-01-04 R. Steven Rainwater <steve@ncc.com>

	* site.c (virgule_site_render_person_link): Removed hardcoded anchor
	style. The external stylesheet can now be used to control appearance
	of the person links.


2006-12-27 R. Steven Rainwater <steve@ncc.com>

	* acct.c (acct_person_graph_serve): Added some sanity checks to the
	graph.dot code that should prevent it from segfaulting if it hits
	a malformed user account XML file.

	* mod_virgule (test_page): Added display of vr->priv->base_uri and
	vr->priv->base_path to test page.
	
	* mod_virgule (virgule_handler): Fixed minor bug in config.xml
	pathname (had extra slash - didn't seem to hurt anything).

	* req.c (virgule_get_tmetric): Fixed bug that was causing trust
	metric cache to be loaded unnecessarily.


2006-12-20 R. Steven Rainwater <steve@ncc.com>

	* acct.c (virgule_acct_person_index_serve): Removed some hard-coded
	HTML that should be in the person index template.

	* article.c: Cleaned up some HTML formatting cruft.

	* diary.c (virgule_diary_store_feed_item): The item's unique 
	identifier is now stored if it's available.

	* tmetric.c (tmetric_test_serve): Removed this unused DB lock test.


2006-12-13 R. Steven Rainwater <steve@ncc.com>

	* aggregator.c (aggregator_index_atom_10): Added support for using 
	the required atom:entry element, atom:id as a unique item identifier
	string.

	* aggregator.c (aggregator_index_rdf_site_summary_10): Added support
	for using the required item attribute, rdf:about, as a unique item
	identifier string.

	* aggregator.c (aggregator_index_rss_20): Unlike Atom or RDF Site
	Summary, RSS has no required unqiue identifier for items. So, we
	first try to use the optional guid element. If that's not present, we
	try to use the link element, which should be unique. If neither is
	present, we make a final attempt to use the title element which
	might be unique (but possibly not stable). RSS sucks.


2006-12-06 R. Steven Rainwater <steve@ncc.com>

	* acct.c (acct_index_serve): Corrected minor HTML glitch on login
	form.


2006-12-06 R. Steven Rainwater <steve@ncc.com>

	* diary.c (virgule_diary_store_feed_item): Improved error checking
	on entryposttime tag when storing feed item in blog.


2006-12-03 R. Steven Rainwater <steve@ncc.com>

	* aggregator.c (aggregator_identify_feed_type): Added a check for
	Atom version 0.3 feeds.

	* aggregator.c (aggregator_normalize_feed): Added case for Atom
	version 0.3 feeds.

	* aggregator.c (aggregator_index_atom_10): Added support for the Atom
	version 0.3 tags issued and modified, which are used in place of the
	newer published and updated tags.

	* aggregator.c (virgule_update_aggregator_list): Normalizes the feed
	URL to include http:// in case the user forgot it.

	* db.c (virgule_db_lock_downgrade): New function. Downgrades a write 
	lock to a shared read lock.

	* aggregator.c (aggregator_post_feed): Made the global write lock
	usage more granular to improve site performance while the aggregator
	is running. The lock is now held only during the actual blog update
	and not during the feed retrieval and parsing process.

	* rating.c (rating_crank_all): The rating loop now releases the XML
	file system lock, sleeps for 1 microsecond, and re-acquires the lock
	on each iteration. This increases the total time taken for the
	eigen vector update slightly but it allows normal use of the site to
	continue during the calculation, eliminating the several minute wait
	at the top of each hour.


2006-12-02 R. Steven Rainwater <steve@ncc.com>

	* util.c (virgule_rfc822_to_time_t): Added check for the quasi-RFC822
	date format generated by pyBlosxom blogs.


2006-11-30 R. Steven Rainwater <steve@ncc.com>

	* aggregator.c (aggregator_post_feed): Made the handlings of updates
	a little smarter so we don't bother trying to update an entry that
	was posted from the same instance of the feed.
	
	* diary.c (virgule_diary_update_feed_item): Added check for NULL
	key for the target entry. When a feed is processed the first time,
	it's possible to find an update for an entry that we don't have yet.


2006-11-29 R. Steven Rainwater <steve@ncc.com>

	* aggregator.c (aggregator_post_feed): Now handles entries that have
	been updated by passing them to virgule_diary_update_feed_item().
	At present only Atom feeds seem to have support for updated entries.

	* diary.c (virgule_diary_update_feed_item): Handles the replacement
	of a diary entry that has been updated or changed via a syndicated
	feed.

	* diary.c (find_entry_by_feedposttime): Searches blog entries in
	reverse chronological order trying to find one that matches the
	specified feed post time.

	* diary.c (virgule_diary_store_feed_item): Stores the update time
	of entries when available. At present, only Atom feeds seem to
	include an entry update time tag.

	* diary.c (virgule_diary_entry_render): Diary update timestamp is
	added to syndication link when entry has been updated via a feed.

	* util.c (virgule_virgule_to_time_t): Converts the timestamp string
	format used by mod_virgule into a proper time_t value, adjusting for
	the server's time zone. The older virgule_iso_to_time_t was broken.
	This code should replace the old code in all uses eventually.

	* xml_util.c (virgule_xml_del_string_contents): New function that
	removes and frees all text and CDATA child nodes from an XML node
	while leaving other children untouched.


2006-11-22 R. Steven Rainwater <steve@ncc.com>

	* aggregator.c (aggregator_index_rss_20): Added support for <guid>
	tag as an alternate blog entry permalink when the <link> tag is
	missing. 

	* diary.c (virgule_diary_entry_render): Created a new function to
	render a diary entry and associated header. This combines a lot of 
	redundant code from several functions in diary.c and site.c, making
	it easier to maintain. Diary/blog entries will now include the entry
	title if it exists. At present only syndicated entries have titles.

	* diary.c (virgule_diary_latest_render): Removed this function. It
	has been entirely replaced by virgule_diary_entry_render().

	* diary.c (virgule_diary_render): Removed duplicate diary rendering 
	code from this function. It now relies on the new function,
	virgule_diary_entry_render(). 

	* site.c (site_render_person_link): Added virgule_ prefix.
	
	* site.c (site_render_recent_changelog): Removed duplicate diary
	rendering code from this function. It now relies on the new function,
	virgule_diary_entry_render().


2006-11-20 R. Steven Rainwater <steve@ncc.com>

	* aggregator.c: Added xmlChar/char casts to fix gcc 4.x warnings.
	* certs.c:  Added xmlChar/char casts to fix gcc 4.x warnings.

	* rss_export.c (rss_render_from_xml): pubDate is now added to RSS 0.91
	feed to make it easier to sort items. The description content is now 
	properly escaped.


2006-11-16 R. Steven Rainwater <steve@ncc.com>

	* util.c (virgule_UTF8ToHtml): This is a local, static copy of the
	UTF8ToHTML function from libxml2. It has a couple of minor changes
	that allow it to handle ALL UTF-8 character correctly rather than
	only UTF-8 characters for which named entity values exist in HTML.
	Numeric entity values are used if a named entity is not available.

	* util.c (virgule_nice_utf8): Now relies on our fixed version of the
	UTF8ToHTML code.

	* util.c (virgule_nice_htext): Now calls virgule_nice_utf8 to make
	sure HTML is in the form of ASCII + entity values or numerical 
	character values before	processing. This should fix the bug with 
	Chinese (and other) character sets not being saved and rendered 
	correctly.


2006-11-15 R. Steven Rainwater <steve@ncc.com>

	* aggregator.c (aggregator_index_atom_10): Now handles missing link
	within items correctly.

        * xml_util.c (virgule_xml_get_string_contents): Bug fix. Now handles
	CDATA sections correctly.

	* acct.c (acct_kill): The aggregator feedlist is checked during
	account deletion and the user's feed is removed if it is present.

	* aggregator.c (aggregator_index_rss_20): Now checks for the presence
	of a content:encoded tag and uses that in place of the description
	tag contents if found. Alternate formats that use dc:date in place
	of pubDate are also supported. 


2006-11-14 R. Steven Rainwater <steve@ncc.com>

	* aggregator.c (aggregator_index_rdf_site_summary_10): RDF Site Summary
	parsing is now working. Only one variant using a combination of the 
	Dublin Core module for dates and a draft version of the Content
	module for content encoding has been tested. Other variants are 
	unlikely to work.


2006-11-13 R. Steven Rainwater <steve@ncc.com>

	* aggregator.c (aggregator_index_rss_20): RSS parsing is now working.
	Most versions should work but older versions will need the optional
	pubDate tag in order to be parsed correctly.

	* acct.c (acct_newsub_serve): Increased recent account listings from
	50 to 100 for admin reasons. It's helpful to keep a month or so of
	listings since some spammers wait for while before doing anything
	interesting.

	* util.c (virgule_time_t_to_iso): Converts a time_t value to an
	virgule-style ISO string. If no time_t value is given, the current
	time is rendered. 

	* util.c (virgule_rfc822_to_time_t): Converts an RFC-822 encoded
	time to a time_t value.


2006-11-12 R. Steven Rainwater <steve@ncc.com>

	* diary.c (virgule_diary_store_feed_item): Stores an item from a 
	syndicated feed in a user's local blog. Several elements are added 
	to the XML document that are not included when a local diary entry 
	is made, such as the permalink to the original (remote) blog post.


2006-11-11 R. Steven Rainwater <steve@ncc.com>

	* aggregator.c (item_compare): A simple qsort function to compare
	the timestamps of blog items.

	* mod_virgule.c (read_site_config): Added computation of the local
	server's time offset from UTC in seconds. We aren't using this yet
	but will need it in the near future.	


2006-11-10 R. Steven Rainwater <steve@ncc.com>

	* aggregator.c (aggregator_normalize_feed): Takes whatever kind of
	feed you've got and returns a normalized structure with normalized
	timestamps, with items sorted by posting time.

	* aggregator.c (aggregator_index_atom_10, aggregator_index_rss_20,
	aggregator_index_rdf_site_summary_10): Functions to normalize and
	index Atom, RSS, and RDF Site Summary feeds. At present we use a
	small subset of the tags in each, so it seems safe enough to use
	these for other versions of their kind (e.g. the rss_20 function
	should work fine on RSS 0.91 feeds, etc.)

        * xml_util.c (virgule_xml_get_string_contents): Verifies that the
	passed xmlNode is non-NULL before using it. May have been the
	source of some segfaults.

	* util.c (virgule_rfc3339_to_time_t): Converts an RFC-3339 encoded
	time to a time_t value.


2006-11-09 R. Steven Rainwater <steve@ncc.com>

	* diary.c (diary_edit_serve): Minor correction to error message
	text. Added missing quotes to attributes in HTML tags.

	* proj.c (proj_proj_serve): Fixed bug that caused all the projects
	to vanish. A missing bracket on an if statement was causing all
	project pages to return a 404.

	* db_ops.c (virgule_remove_recent): Fixed bug that limited removal
	of entries from recent files to the initial instance. Now all 
	instances of an entry will be removed.
	

2006-11-07 R. Steven Rainwater <steve@ncc.com>

	* aggregator.c (aggregator_identify_feed_type): Attempts to identify
	the format of a blog feed by examining the XML file. Initially it
	works on most versions of Atom, RSS, and RDF Site Summary formats.


2006-11-05 R. Steven Rainwater <steve@ncc.com>

	* diary.c (virgule_diary_latest_feed_entry): Returns a Unix time_t 
	value reflecting the timestamp of the most recent syndicated diary 
	entry for the specified user.

	* aggregator.c (aggregator_post_feed): Attempts to identify new blog
	posts in the user's feed buffer and appends them to the user's blog.


2006-11-03 R. Steven Rainwater <steve@ncc.com>

	* aggregator.c (virgule_aggregator_serve, aggregator_getfeeds_serve):
	When /aggregator/getfeeds.html is hit, the feedlist will be opened
	and each listed feed will be retrieved, buffered, and processed.


2006-11-02 R. Steven Rainwater <steve@ncc.com>

	* aggregator.c (virgule_update_aggregator_list): Created this function
	to check the specified user account and add them to or delete them 
	from the aggregation list depending on their current account settings.


2006-11-01 R. Steven Rainwater <steve@ncc.com>

	* priv.h: Added a user blog syndication level that can be set from 
	the config.xml file. This level will determine what certification
	is needed for a user to access the blog syndication feature.

        * mod_virgule.c (read_site_config): Now reads the blogsyndicate
	feature access level.

	* req.c (virgule_ok_to_syndicate_blog): New function that returns
	TRUE if the specified user is certified at a level high enough to
	use the blog syndication features. Result is based on the current
	setting specified in the config.xml file.

	* acct_maint.c (account_index_serve, account_update_serve): Users
	can now edit and save blog aggregation info to profile.xml. Syndicate
	status is forced to OFF for users who are not certified at a level 
	allowed to use syndication.


2006-10-31 R. Steven Rainwater <steve@ncc.com>

        * priv.h: Added pointer to tmetric data in the Apache thread private
	pool. This will allow us to load tmetric data only when it changes
	and preserve it between hits, instead of loading on each hit. 
	A pointer to a memory sub pool and a time stamp for the tmetric
	cache was also added.

	* mod_virgule.c (read_site_config): Make sure the thread private
	tmetric pointer is set to NULL when the Apache process initializes.
	The tmetric cache sub pool is destroyed prior to destruction of the
	main thread private pool when a config file reload occurs.

	* req.c (virgule_req_get_tmetric): A stat is done on the tmetric
	cache and it's only reloaded if it has been updated since the last
	load. Otherwise, a pointer to the copy already loaded in the thread 
	private memory pool is return. If a tmetric reload occurs, the
	tmetric sub pool is destroyed and a new sub pool allocated. This
	should prevent memory leaks on reload.

	* tmetric.c (virgule_tmetric_get): The trust metric cache is now
	loaded using memory allocated from the tmetric sub pool.

	* proj.c (proj_proj_serve): The "Project not found" page now returns
	a 404 instead of a 200. This should discourage repeated visits from
	search engines to dead project pages.
	
	* util.c (virgule_nice_utf8): Added xmlChar casts to eliminate gcc 4.x
	warning. Replaced strlen with xmlStrlen.

	* xml_util.c: Added xmlChar/char casts to eliminate gcc 4.x warnings.

	* acct_maint.c (acct_person_serve): Added a "member since" field to 
	the user profile page.


2006-10-25 R. Steven Rainwater <steve@ncc.com>

	* auth.c (virgule_auth_user_with_cookie): Added xmlChar/char casts
	to eliminate gcc 4.x warnings.

	* article.c (article_generic_submit_serve): Added xmlChar/char casts
	to eliminate gcc 4.x warnings.

	
2006-10-22 R. Steven Rainwater <steve@ncc.com>

	* acct_maint.c: Fixed numerous xmlChar/char castings to fix gcc 4.x
	warnings. Switched a few strcmp() calls to xmlStrcmp() both to fix
	warnings and to make sure the string handling code is UTF-8 safe.


2006-10-19 R. Steven Rainwater <steve@ncc.com>

	* Makefile: Minor compatibility changes for gcc 4.x and Apache 2.2.

	* mod_virgule (command_rec table): Updated to remove reliance on
	deprecated Apache 1.3 compatibility layer. Should compile cleanly
	with Apache 2.2 as well as 2.0 now.
	
	* mod_virgule (read_site_config): Added several xmlChar casts of
	char to	eliminate gcc 4.x warnings. Also replaced calls to strcmp
	with calls to xmlStrcmp.

	* net_flow.c (virgule_net_flow_max_flow): #Ifdef'd out several
	debug g_print statements.

	* eigen.c (virgule_eigen_crank): Added a few xmlChar and char
	casts to eliminate gcc 4.x warnings.

	* acct_maint.c (acct_person_serve): When a "person not found" page
	is returned, the HTTP status code is set to 404 instead of 200. This
	should help search engines to stop requesting dead pages.
	
	* acct_maint.c (acct_person_diary_serve): When diary entries for a
	non-existent person are requested, the HTTP status code is set to
	404 and a "person not found" error page is returned.

	* tmetric.c (virgule_tmetric_get): Removed some debugging cruft.
	

2006-10-16 R. Steven Rainwater <steve@ncc.com>

	* acct_maint.c (acct_index_serve): Separated login and password
	reminder forms to improve usability.

	* acct_maint.c (acct_person_serve): Spam reporting is now done through
	a form as per Scott Lamb's suggestion. Helps protect against accidental
	reports by automated agents following links. This also keeps us
	compliant with RFC 2616.
	

2006-10-13 R. Steven Rainwater <steve@ncc.com>

	* rating.c (rating_crank_all): Rewrote this function completely. It
	now uses the cached, pre-sorted tmetric list instead of looping
	through and reading every account's XML file. This shaved about 30%
	off the total processing time.
	

2006-10-09 R. Steven Rainwater <steve@ncc.com>

	* mod_virgule (read_site_config): Added config handler for new
	<accountspamthreshold> tag to set the spam score at which an
	untrusted observer account is automatically deleted.

	* acct_maint.c (acct_flag_as_spam): Handles an account spam report. 
	Writes a new spam flag to the reported account with the name of the 
	flag issuer, and recalculates the spam score. If the score exceeds
	the currently configured threshold, the account is removed.

	* acct_maint.c (virgule_acct_person_index_serve): User list paging
	now handles missing accounts smoothly (most likely missing because
	they got deleted by the new acct_kill function).

	* eigen.c (virgule_eigen_cleanup): Remove any eigen data associated
	with the passed user account. Called as part of the account deletion
	process.

	* db_ops.c (virgule_remove_recent): Remove items from a recent list
	that match a specified user account name. Called as part of the
	account deletion process.


2006-10-08 R. Steven Rainwater <steve@ncc.com>

	* acct_maint.c (acct_person_serve): Now displays the spam rating for
	observer accounts. A spam report link is displayed on observer 
	accounts when viewed by trusted users.

	* req.c (virgule_req_get_tmetric_level): Added additional sanity
	check to handle case where vr->u is null. This was causing a
	segfault when a non-trusted user performed certain (very rare)
	operations.


2006-10-06 R. Steven Rainwater <steve@ncc.com>

	* tmetric.c (tmetric_index_serve): Removed unneeded HTML dump of
	trust metric debug data during processing. This reduced trust
	metric computation time by more than 50%.


2006-10-05 R. Steven Rainwater <steve@ncc.com>

	* acct_maint.c (acct_index_serve): Corrected typo in UTF-8 content
	type specifier for account info form.

	* acct_maint.c (acct_kill, acct_killsub_serve): Updated account 
	removal code for live use. Account kill feature is now live and can 
	be used by any of the users listed as special users in config.xml.

	* article.c (article_render_from_xml): Added missing UTF-8 content
	type specifier in reply form. 

	* article.c (article_reply_submit_serve): Added code to (hopefully) 
	reject duplicate replies.


2006-10-04 R. Steven Rainwater <steve@ncc.com>

	* util.c (virgule_user_is_special): Tests whether or not the currently
	logged in user, if any, is on the list of special (admin) users.
	
	* article.c (article_render_from_xml): Replaced special user code
	with call to virgule_user_is_special().

	* cert.c (virgule_render_cert_level_begin): Replaced special user
	code with all to virgule_user_is_special().
	

2006-10-03 R. Steven Rainwater <steve@ncc.com>

	* acct_maint.c (send_email): For added security the IP address 
	requesting a password reminder is included in the reminder email.

	* acct_maint.c (acct_person_serve): Observers now have a nofollow
	relation attribute added to the anchor tag for their homepage.

	* util.c (virgule_add_nofollow): Adds nofollow relationship
	attributes to any anchor tags in the passed string.

	* diary.c (virgule_diary_render, virgule_diary_latest_render): 
	Nofollow is added to anchors in	diary entries when rendered for an
	observer.


2006-10-02 R. Steven Rainwater <steve@ncc.com>

	* article.c (article_render_from_xml): Reply posting form is now
	concatenated to end of reply list as per Advogato style.
	
	* util.c (virgule_nice_utf8): Added new function to clean up some
	UTF-8 to HTML entity conversions that were being missed.
	

2006-10-01 R. Steven Rainwater <steve@ncc.com>

	* mod_virgule.c (virgule_read_site_config): Added code support for
	<articletitlesize> tag in config.xml that can set the maximum length
	allowed for article titles.
	

2006-09-29 R. Steven Rainwater <steve@ncc.com>

	* proj.c: Now has even more horrible hacks scattered through it than
	before to render three "styles" of project pages: RAPH, NICK, and
	STEVE. The project style can be set in the config file, so provided
	you can use one of the three existing options, no code changes will be
	needed to customize mod_virgule for your use. RAPH style projects match
	the software projects of Advogato. STEVE style is identical except for 
	one field name change that makes project more generic and matches the
	use on robots.net. NICK style converts the project area into a sort of
	makeshift discussion forum. This whole file is a disaster and needs to
	be re-written to use an external schema or replaced with a general 
	purpose XML object handler.

	* schema.c:  Added virgule_ prefix to non-static functions.
	* site.c:  Added virgule_ prefix to non-static functions.
	* style.c:  Added virgule_ prefix to non-static functions.
	* tmetric.c:  Added virgule_ prefix to non-static functions.
	* util.c:  Added virgule_ prefix to non-static functions.
	* wiki.c:  Added virgule_ prefix to non-static functions.
	* xml_util.c:  Added virgule_ prefix to non-static functions.
	* xmlrpc.c:  Added virgule_ prefix to non-static functions.


2006-09-28 R. Steven Rainwater <steve@ncc.com>

	* mod_virgule.c (virgule_read_site_config): Added code support for
	three settings to determine the features available at particular
	certification levels including <articlepost>, <articlereply>, and
	<projectcreate>. 

	* req.c (req_ok_to_post, req_ok_to_reply, req_ok_to_create_project):
	These functions now return TRUE or FALSE based on the cert levels
	set in the config.xml file.

	* article.c (article_render_from_xml): Fix to account for special
	users in CSS setting of article headers (to match original Advogato
	functionality).
	
	* style.c (render_sitemap): Anchor tags are supressed on the sitemap
	map link if the target URI matches the URI of the current page. 

	* req.c:  Added virgule_ prefix to non-static functions.
	* rss_export.c:  Added virgule_ prefix to non-static functions.


2006-09-23 R. Steven Rainwater <steve@ncc.com>

	* general: Replaced "weblog" with "blog". Robots.net used "weblog"
	and Advogato used "diary". After discussing with Raph, we decided
	"blog" was the most recognizable choice.

	* hashtable.c: Added virgule_ prefix to non-static functions.
	* net_flow.c: Added virgule_ prefix to non-static functions.
	* proj.c:  Added virgule_ prefix to non-static functions.
	* rating.c:  Added virgule_ prefix to non-static functions.


2006-09-21 R. Steven Rainwater <steve@ncc.com>

	* mod_virgule.c (virgule_read_site_config): Added code support for
	<accountextendedcharset> tag in config.xml. When set to "on", this 
	will allow new account name to use a slightly more premissive set of
	characters including -,_,., and spaces in their names. When set
	to "off" it uses the original mod_virgule/Advogato charset.

	* acct_maint.c (virgule_validate_username): The allowed character
	set for new user names is determined by a config.xml setting.

	* eigen.c: Added virgule_ prefix to non-static functions.


2006-09-20 R. Steven Rainwater <steve@ncc.com>

	* mod_virgule.c (set_virgule_db): Stats the provided path to verify
	that it exists and that it's a directory. Returns error message to
	Apache if path is invalid.

	* mod_virgule.c (virgule_read_site_config): Added code support for
	<articletitlelinks> tag in config.xml. When set to "on", this will
	turn article titles into links to the article detail page. When set
	to "off" it uses the original mod_virgule/Advogato style of titles.
	
	* article.c (article_render_from_xml): Article title link is now
	dependent on config.xml setting. Article header white space and
	width behavior is now set in CSS rather than hardcoded HTML.

	* style.c (render_header): Isolated a nasty chunk of hardcoded
	HTML and created a generic version that will be used on all sites
	except robots.net. This is a particulary nasty hack and will be 
	replaced by something cleaner later.
	
	* diary.c: Added virgule_ prefix to non-static functions.

2006-09-18 R. Steven Rainwater <steve@ncc.com>

	* db.c: Added virgule_ prefix to non-static functions.
	* db_obs.c: Added virgule_ prefix to non-static functions.
	* db_xml.c: Added virgule_ prefix to non-static functions.

2006-07-07 R. Steven Rainwater <steve@ncc.com>

	* acct_maint.c: Added virgule_ prefix to non-static functions.
	* apache_util.c: Added virgule_ prefix to non-static functions.
	* article.c: Added virgule_ prefix to non-static functions.
	* auth.c:  Added virgule_ prefix to non-static functions.
	* buffer.c:  Added virgule_ prefix to non-static functions.
	* certs.c:  Added virgule_ prefix to non-static functions.

2006-06-13 R. Steven Rainwater <steve@ncc.com>

	* util.c (b64enc): Added static to function definition.
	* util.c (rand_cookie): Renamed virgule_rand_cookie.
	* util.c (add_topic): Renamed virgule_add_topic.
	* util.c (match_preifx): Renamed virgule_match_prefix.
	* diary.c (diary_serve): Renamed virgule_diary_serve.
	* mod_virgule.c (read_site_config): Renamed virgule_read_site_config.
	* proj.c (proj_serve): Renamed virgule_proj_serve.
	* rating.c (rating_serve): Renamed virgule_rating_serve.
	* rss_export.c (rss_serve): Renamed virgule_rss_serve.

2006-03-14 R. Steven Rainwater <steve@ncc.com>

	* article.c (article_num_serve): Rewrote to use a user editable
	XML template populated by site_render_page() instead of a
	hard-coded page. All article non-form pages are now fully user 
	configurable.
	
2006-02-23 R. Steven Rainwater <steve@ncc.com>

	* article.c (article_num_serve): Removed some debugging cruft
	
	* site.c (_RenderCtx): Added variables to hold a tag string and
	buffer string. This will allow site_render to populate rendered XML
	templates with contents rendered by other mod_virgule functions. 
	
	* site.c (site_render_page): Added tag and buffer string args that
	are passed to site_render via the ctx variable. This function can 
	now be called directly from anywhere in mod_virgule to render a 
	template and optionally populate it with locally rendered content.

2006-02-17 R. Steven Rainwater <steve@ncc.com>

	* mod_virgule.c (register_hooks): Renamed to virgule_register_hooks
	to comply with Apache module definition standards.
	
	* mod_virgule.c (create_virgule_dir_conf): Renamed to
	virgule_create_dir_config to comply with Apache module definition
	standards.

	* mod_virgule.c (read_site_config): Added additional memory allocation
	safety checks.

	* acct_maint.c (acct_person_serve): Added last login date to the info
	displayed on the account page.
	
	* db_ops.c (add_recent, db_relation_put_field): Added additional null
	pointer checks.

2005-08-29 R. Steven Rainwater <steve@ncc.com>

	* site.c (site_render, site_render_cronbox, site_render_recent_bots):
	Removed the cronbox and recentbot tags and associated render
	functions as both of these have been replaced by the more general
	include tag.
	
2005-08-26 R. Steven Rainwater <steve@ncc.com>

	* mod_virgule.c (read_site_config): Added code to read a new 
	user configurable admin email address tag, adminemail, from the
	site config.xml

	* acct_maint (send_email): Replace hard-coded From: address on
	password reminder email with the configurable admin address.
		
	* article.c (article_render_recent): Replace hard-coded story
	suggestion email address with the configurable admin address.

2005-08-25 R. Steven Rainwater <steve@ncc.com>

	* mod_virgule.c (test_page, read_site_config): Added support for 
	user configurable section of config.xml for article topics.

	* private.h: Added use_article_topics variable. Added topic array
	header. Added topics to point to the list of Topic structs.

	* req.h: Removed old topic array header

	* util.c (add_topic): New function to add topics as they are
	read from the config file.

	* article.c (lots-o-functions): Replaced #ifdefs and hard-coded
	support for topics with support for user-configurable topic setting.

2005-08-19 R. Steven Rainwater <steve@ncc.com>

	* site.c (site_render, site_render_page): Implemented the
	head_content tag. Head contents are now passed to the 
	render_header() function where they are inserted into the 
	rendered HTML.
	
2005-07-27 R. Steven Rainwater <steve@ncc.com>

	* Makefile: Added --strip-debug to linker options to keep filesize
	reasonable. This is done already on some distros depending on how
	they build Apache. Explicitly adding the option just makes sure it
	gets done consistently on all builds.
	
2005-07-25 R. Steven Rainwater <steve@ncc.com>

	* certs.c (render_cert_level_begin, render_cert_level_end): Removed
	unneed calls to render_table_open() and render_table_close().

	* site.c (render_footer): Removed this redundant function. The
	function render_footer_send now does it all.

	* site.c (site_render): Removed unneeded calls to render_table_open()
	and render_table_close().

	* style.c (render_header_raw, render_footer): Removed unneeded STYLE
	ifdefs and related code used to render old-style <FONT> tags.
	
	* style.c (render_table_open, render_table_close): Removed these
	unused functions.
	
	* style.c (renders_sitemap): Removed unneeded STYLE ifdefs while
	retaining the enclosed flag for advogato compatiblity.

	* general: Added a sample data/intermap.txt file to the sample_db
	directory so that wiki tags will work on new installs.

2005-07-22 R. Steven Rainwater <steve@ncc.com>

	* diary.c (diary_rss_export): Fixed segfault that occured if an
	RSS XML file was built from an empty diary entry.

2005-07-20 R. Steven Rainwater <steve@ncc.com>

	* Makefile: Added -fno-strict-aliasing flag to get around the
	type-punned pointer problem in eigen.c that prevents compilation
	by gcc 3.4.3.
	
2005-07-10 R. Steven Rainwater <steve@ncc.com>

	* private.h: Created an include for mod_virgule thread private data
	structures. Moved some elements of the VirguleReq data structure
	found in req.h to a virgule_private data structure in this file.

	* (general): Modified the site configuration handling to use a thread
	private data pool that persists across requests. The site config.xml
	file is now loaded into virgule_private once for each Apache thread
	instead of into req once for each request. If the config.xml file is
	updated, it will be reloaded as needed.

	* mod_virgule.c (test_page): Now displays status of the new account
	creation setting of the configuration. The time stamp of the config
	file in use is also displayed.
	
2005-07-01 R. Steven Rainwater <steve@ncc.com>

	* (general): Diaries are known as weblogs today by just about everyone
	and it seems to confuse users to refer to them as anything else. So
	I'm updating all textual references throughout the code from "diary"
	to "weblog". For now, I'm leaving "diary" in the underlying code and
	URL references but that should probably be modernized at some point.

2005-06-22 R. Steven Rainwater <steve@ncc.com>

	* acct_maint.c (acct_person_serve): Fixed a very old HTML coding
	error - a missing close tag on the #cert fragment identifier.
	
2005-06-16 R. Steven Rainwater <steve@ncc.com>

	* db.c (db_get_p): A segfault was caused if db_get_p was called with
	a NULL key. It now handles this condition gracefully by returning a
	NULL result. This was causing occasional segfaults for trustmetric
	computations on user profiles that had damaged or incorrect names.
	
	* tmetric (tmetric_run): The main loop skips files which return a
	NULL db_key value. This is usually caused by a damaged or incorrect
	profile name. Previously, the NULL key value was passed to db_get_p()
	resulting in a segfault.

2005-06-12 R. Steven Rainwater <steve@ncc.com>

	* (general): Ported mod_virgule to the Apache 2 module API. Changes
	to includes and code in all *.c files were required.

2005-02-27 R. Steven Rainwater <steve@ncc.com>

	* acct_maint.c (acct_person_serve): Punched up the wording below the
	certify form to try to cut down on incorrect or inflated certs.
	
2005-01-29 R. Steven Rainwater <steve@ncc.com>

	* site.c (site_render): Added check for a "start" argument on pages
	that use <articles> tag to allow paging of an external template.
	
	* articles.c (article_older_serve): Removed. No longer needed now
	that article_recent_render() handles paging correctly. Also, note
	that the site/article/index.xml and site/article/older.xml template
	are essentially identical now for stock mod_virgule sites. The only
	difference is that index.xml starts with the most recent articles
	by default and older.xml starts one page back. This was done because
	the user may wish to customize the two page to be slightly different
	in some cases.

	* style.c (dayofweek): Added a function that returns the day of week
	so that we can do full RFC 822 dates as needed by RSS 2.0.

	* style.c (render_date): Added a third date format for full RFC 822
	date strings.

	* rss_export.c (rss_serve, rss_render, rss_render_from_xml): Added 
	RSS 2.0 support to the existing RSS 0.91 code. The URLs are:

	RSS 0.91 - /rss/articles.xml
	RSS 2.0  - /rss/articles-2.0.xml
		
2004-10-22 R. Steven Rainwater <steve@ncc.com>

	* site.c (site_render): Added a <userlist> tag that will place a 
	pageable list of mod_virgule users on a page. Next/previous page
	navigation is automatically added and there is a max-per-page value
	that may be set in the tag to determine how many users are listed
	on each page.
	
	* acct_maint.c (acct_person_index_serve): Complete rewrite of this
	function to make it much faster, allowing incremental display and
	paging. Instead of completely parsing the user accounts and sorting
	them in memory, the userlist is now read from the pre-sorted trust
	metric table and only those accounts on the currently displayed
	page are loaded and parsed.

	* tmetric.c (node_info_compare): Added the user cert level to the
	comparison. Users are now sorted first by cert level, then by name.
	
2004-10-10 R. Steven Rainwater <steve@ncc.com>

	* db.c (db_put_p): Reverted Gary's diskspace patch yet again. This
	time due to what appears to be a rarely encountered race condition
	that results in the complete loss of the file being written. Yikes!

2004-09-26 R. Steven Rainwater <steve@ncc.com>

	* db.c (db_put_p): Added check for existence of record before rename
	to prevent errors on posting with Gary's disk space patch. The patch
	should now working as intended.
	
	* post.c, post.h: Remove these two files which are not referenced in
	the makefile and appear to be unused cruft.
	
2004-09-17 R. Steven Rainwater <steve@ncc.com>

	* article.c (article_index_serve): Removed this unnecessary function
	and replaced it with a default XML file that makes the page easily
	configurable by the user.

	* site.c (site_render_include): Added a function that allows the user
	to include one XML file within another. 

2004-08-19 R. Steven Rainwater <steve@ncc.com>

	* db.c (db_put_p): Reverted Gary's disk space patch due to problems
	with posting.

2004-08-19 R. Steven Rainwater <steve@ncc.com>

	* db.c (db_put_p): Added Gary Benson's patch to allow graceful 
	failure in case disk space is exhausted.

	* rec.h (_VirguleRec): Added nav_options, and pointer to a list of
	navigation options from which the sitemap can be constructed.

	* util.c (add_nav_option): Added a utility function to create
	sitemap navigation option structures.
	
	* mod_virgule.c (read_site_config): Added code to read the sitemap
	navigation options from the XML config file.

	* style.c (render_sitemap): Sitemap dynamically renders menu from
	the navigation options specified in the XML config file rather than
	using hardcoded values.
	
	* style.c (render_site_link): Removed this function. It is no longer
	needed now that relative paths for the sitemap options may be 
	specified directly in the XML config file.
	
2004-07-09 R. Steven Rainwater <steve@ncc.com>

	* tmetric.c (tmetric_set_name): Fixed segfault that occured if
	tmetric_find_node() was unable to find a user name.
	
	* acct_maint.c (acct_update_serve, acct_person_serve,
	acct_certify_serve): Added Gary Benson's page refresh patches.

2004-06-23 R. Steven Rainwater <steve@ncc.com>

	* acct_maint.c (acct_index_serve): Added missing vr->prefix to the
	"edit your existing projects" link, pointed out by Seth Cohn.
	
2004-04-18 R. Steven Rainwater <steve@ncc.com>

	* (general): Merged changes from latest version of Raph's mod_virgule
	(which is still going by v1.41) including the new account creation
	switch, the XML-RPC check cookie method, recentlog permalinks, article
	permalinks, project permalinks, and recentlog edit links.
	
	* acct_maint.c (acct_index_serve, acct_newsub_serve): Added "forgot 
	password" checkbox and instructions to login form. Added forgotten
	password handler to newsub_serve.
	
	* acct_maint.c (send_email): Added send email function to support the
	"forgot password" feature. This function and the other modifications
	related to the forgotten password feature are based on a patch by
	Steve Kemp <steve@steve.org.uk>. The send_email function has a 
	hard-coded path to sendmail and a hard-coded "from:" value that will
	need to be changed to match the site owners email address. The -f
	flag is used on sendmail, so you'll want to make apache a trusted
	user to avoid authentication warnings in the emails.

	
2003-11-14 R. Steven Rainwater <steve@ncc.com>

	* acct_maint.c (acct_person_index_serve): Corrupt profile.xml files
	should now be silently skipped rather than segfaulting Apache.
	
2003-11-02 R. Steven Rainwater <steve@ncc.com>

	* (general): Changed encoding type from ISO-8859-1 to UTF-8 throughout
	code. This may eliminate some case of browsers submitting forms in
	the wrong encoding.

	* tmetric.c (tmetric_run): Added error checking to prevent attempted
	dereferencing of a null pointer that occured when reading a corrupt
	profile.xml file. Corrupt files are now silently skipped instead of
	causing Apache to segfault.
	
2003-10-19 R. Steven Rainwater <steve@ncc.com>

	* rss_export.c (rss_index_serve): Applied James Henstridge's patch
	to add missing content type to the returned document.

	* (general): Applied James Henstridge's patch that ads a link element
	for auto-discovery of the diary RSS feeds. This changes the number
	of arguments passed to render_header_raw() and render_header() in
	style.c and affect all calls to those functions throughout the code.

	* style.c (render_header_raw): Added link element for favicon. It's
	hardcoded to be /images/favicon.ico. Probably should make it a
	configuration option later on.

	* db_xml.c (db_xml_put): Added xmlIndentTreeOutput flag to make sure
	libxml2 was indenting and adding newlines to output. This is another
	compatibility behavior and can be removed in the future if newlines
	are added to the xml trees.

2003-10-04 R. Steven Rainwater <steve@ncc.com>

	* acct_maint.c (acct_person_diary_???_serve): Replaced 
	xmlDocDumpMemory() with xmlDocDumpFormatMemory() to provide 
	newlines and indenting in output. This simulates the behavior of
	libxml with respect to whitespace. Ideally these calls should
	eventually be replaced with nodes that implement the desired
	newlines and indentation directly.

	* db_xml.c (db_xml_out): Replaced xmlDocDumpMemory() with
	xmlDocDumpFormatMemory() to provide newlines and indenting in output.

	* rss_export.c (rss_index_serve): Replaced xmlDocDumpMemory() with
	xmlDocDumpFormatMemory() to provide newlines and indenting in output.

	* xmlrpc.c (xmlrpc_create_response): Replaced xmlDocDumpMemory() with
	xmlDocDumpFormatMemory() to provide newlines and indenting in output.

2003-10-03 R. Steven Rainwater <steve@ncc.com>

	* db.c (db_is_dir): Removed superfluous strlen() call.
	
2003-10-01 R. Steven Rainwater <steve@ncc.com>

	* mod_virgule.c (virgule_handler): Added xmlKeepBlanksDefault(0) to
	force libxml2 to supress blank nodes. This allows the existing
	mod_virgule XML code to function properly with libxml2 until it
	can be updated. This can be removed once all xml code is adjusted
	to handle blank nodes correctly.
	
	* (general): Merged changes from latest version of Raph's mod_virgule
	(which is still going by v1.41) including James Henstridge's diary
	patches and coderman's XML-RPC patches.

	* (general): Merged Martijn van Beers <martijn@eekeek.org> libxml2 
	patch to bring mod_virgule up to the current XML library version.

	* style.c (render_header_raw, render_footer): Removed code that added
	redundent copy of page title on every page.

2003-04-21 R. Steven Rainwater <steve@ncc.com>

	* (general): Merged changes from official mod_virgule versions
	1.38 through 1.41
	
2002-09-14 R. Steven Rainwater <steve@ncc.com>

	* style.c (render_sitemap): Added "About" to site menu. This is
	temporary until a configurable menu is available.

2002-04-29 R. Steven Rainwater <steve@ncc.com>

	* (general): Merged changes from official mod_virgule version 1.36
	and 1.37.

2002-04-26 R. Steven Rainwater <steve@ncc.com>

	* (general): Merged changes from official mod_virgule version 1.33,
	1.34, and 1.35. 

2002-04-06 R. Steven Rainwater <steve@ncc.com>

	* buffer.c/h (buffer_new, buffer_write, buffer_send_response):
	Modified functions to send ETags. Note that the calling semantics of
	buffer_send_response have changed. Also generates a Content-MD5 tag,
	since it costs nothing.

	* certs.c/h (render_cert_level_text): Added new function to render
	the certification level as hidden text which shows up in text mode
	browsers but is hidden in graphic browsers.

	* acct_maint.c (acct_login): respects new semantics of the
	modified buffer_send_response function.
	
	* article.c (article_render_reply): Added call to the new
	render_cert_level_text functions.
	
	* mod_virgule (virgule_init_handler): Added version number and an
	init_handler to pass it back to Apache.
	
	* req.c (send_response): respects new semantics of the
	modified buffer_send_response function.
	
	* sites.c (site_render_recent_acct, site_render_recent_changelog):
	Added calls to new render_cert_level_text function.

	* xmlrpc.c/h, xmlrpc-methods.c/h: Brought all XML-RPC functions up
	to mod_virgule 1.31 level.

2002-03-26 R. Steven Rainwater <steve@ncc.com>

	* article.c (article_form_serve, article_generic_submit_serve):
	Changed height of lead box to 6 rows from 4. Added link to a
	posting guidelines page.
	
2002-03-24 R. Steven Rainwater <steve@ncc.com>

	* tmetric.c (tmetric_index_serve): Escapes space chars in username 
	before writing the default cert file. This particular bug has been 
	around a long time but only surfaces if there are two users with 
	similar names such as "Joe" and "Joe Smith" in which case, the 
	wrong cert if found.
	* req.c (req_get_tmetric_level): Escapes space chars in username.
	
2002-03-22 R. Steven Rainwater <steve@ncc.com>

	* Merged all changes from Raph's mod_virgule through 2.27 except
	hardcoded directory exception (we already have configurable ones).
	
	* xmlrpc-methods.c/h, xmlrpc.c/h: Added Gary Benson's XML-RPC API.
	
	* auth.c/h (auth_user): Gary Benson's XML-RPC API. Moved bulk of 
	auth_user() into new function auth_user_with_cookie().

	* acct_maint.c (acct_loginsub_serve): Gary Benson's XML-RPC API.
	Moved bulk of acct_loginsub_serve() into new function acct_login().

	* db.c (db_dir_max_in_level): Added Raph's fix for a bug that caused
	miscounting of numeric directories when the _00.._99 entries follow
	_a.._d in the readdir() order.

	* diary.c/h (diary_post_serve): Gary Benson's XML-RPC API. Moved
	bulk of entry storage code into new function, diary_store_entry().

	* mod_virgule.c (virgule_handler): Gary Benson's XML-RPC API. Added
	handler for xmlrpc stuff.
	
	* tmetric.c (tmetric_index_serve): Added Raph's fix to reduced time 
	spent holding global db lock when writing tmetric cache.

2002-03-18 R. Steven Rainwater <steve@ncc.com>

	* acct_maint.c (acct_kill): Added acct_kill() functio to totally
	remove a user account. The removal procedure will remove all certs
	to and from the user, remove all diary entries, and remove all the
	user's project relations. Projects themselves are not removed (even
	if the user being removed created a project, it may be in use by
	others) so there is a possibility of orphaned projects. I haven't
	decided on the best way to deal with this but it's a small price
	to pay to finally be able to delete user accounts.

	* cert.c (cert_set): Modified to actually remove certs if they are
	set to CERT_LEVEL_NONE instead of just rendering them invisible.
	
	* db.c (db_del): Added db_del() function to remove database records
	by key (i.e., it deletes the associated files). This is the missing
	link that should allow us to eventually add functions that delete 
	user accounts, certs, projects, etc. If removal of the key leaves
	the directory it was residing in empty, the directory is removed too.
	
	* proj.c (proj_set_relation): Added proj_set_relation() function to
	allow procedural changes to a user's relation to a project. This was
	previously possible only through the web interface. This function
	also deletes the relation entirely if it is set to "None" rather than
	just hiding the stale data as the previous implementation.
	
	* proj.c (proj_relsub_serve): Now calls the new proj_set_relation()
	function rather than directly calling db_relation_put().
	
2002-02-17 R. Steven Rainwater <steve@ncc.com>

	* acct_maint.c (acct_person_diary_xml_serve): Added patch from Gary
	Benson to fix memory leak in XML code.
	
	* rss_export.c (rss_index_serve): Added patch from Gary Benson to
	fix memory leak in XML code.
	
2002-02-04 R. Steven Rainwater <steve@ncc.com>

	* rss_export.c (rss_index_serve): Adjust URL and increased number
	of exported articles to 15 from 10.

2002-02-01 R. Steven Rainwater <steve@ncc.com>

	* article.c (article_submit_serve): Added code to authorize user and
	check for duplicate postings (due to user error or browser problems).

2001-08-11 R. Steven Rainwater <steve@ncc.com>

	* style.c (render_header_raw): Added a hard-coded keyword meta tag
	for robots.net.
	
2001-07-18 R. Steven Rainwater <steve@ncc.com>

	* site.c (site_render_banner_ad): Rewrote banner ad code to use the
	XML database. Banner ads are randomly selected from /ads/adlist.xml
	and page view counts are decremented with each viewing of the ad.
	When view count reaches zero, ads are removed from rotation.

	* site.c (site_send_banner_ad): A wrapper for site_render_banner_ad
	that allows banner ad requests used as an SSI within an HTML page.

	* mod_virgule.c (virgule_handler): Added code to handle requests
	for '/cgi-bin/ad' by calling site_send_banner_ad.
	
2001-06-15 R. Steven Rainwater <steve@ncc.com>

	* acct_maint.c (acct_touch): New function. Touches a <lastlogin>
	tag in the profile that can be used by an external account reaper
	to remove unused/unwanted accounts. When I get around to adding
	an internal account reaper, it will also use this function.

	* auth.c (auth_user): Added a call to acct_touch().

	* article.c (article_form_serve): Removed redundant "Post a new
	article" text from Posting page.
	
2001-05-21 R. Steven Rainwater <steve@ncc.com>

	* tmetric.c (tmetric_index_serve): Added changes from  Raph's 1.22
	version of mod_virgule. Neale Pickett's fix was added to the tmetric 
	cache for names preceding the "-".
	
2001-04-27 R. Steven Rainwater <steve@ncc.com>

	* site.c (site_render): Added a tag to for the loginbox.
	
	* site.c (site_render_loginbox): Added this function to display a 
	login box for users who are not logged in or a welcome message for 
	those who are.
	
2001-04-23 R. Steven Rainwater <steve@ncc.com>

	* article.c (article_submit_serve): Added maxlength attributes to
	article submission form to enforce length limits on article titles.
	
	* article.c (article_generic_submit_serve): Added maxlength attributes.
	
2001-04-12 R. Steven Rainwater <steve@ncc.com>

	* article.c (article_recent_render): Added a link for users who can't
	post that allows them to email story suggestions
	
2001-04-11 R. Steven Rainwater <steve@ncc.com>

	* style.c (render_header_raw): Added a line of javascript to prevent
	pages from being displayed inside other sites framesets - this can
	happen from bad HTML on other websites or intentionally to wrap 
	banner ads around someone elses sites.
	
2001-04-06 R. Steven Rainwater <steve@ncc.com>

	* mod_virgule.c (xlat_handler): Added a new handler to accept a
	list of directories in httpd.conf that will be passed through to
	Apache to be handled normally.
	
	* mod_virgule.c (test_page): Added a dump of any VirgulePass values
	
	* mod_virgule.c (set_virgule_pass): New function that stores the
	VirgulePass values in an array.
	
	* mod_virgule.c (set_virgule_dir_conf): Added initializer for the
	new pass_dirs element.

	* site.c (site_render_recent_bots): Added a function to display
	recently added robots on the robomenu from an xml file created by
	the Perl administration program. 
	
	* site.c (site_render): Added check for the <recentbots/> tag.

	* style.c (render_sitemap): Added link to robomenu and changed the
	"person" menu to read "humans".
	
2001-04-03 R. Steven Rainwater <steve@ncc.com>

	* auth.c (auth_user): Tweaked the Apache user name data to remove
	spaces. This appears to be perfectly legal as far as the common log
	format and Apache are concerned but webalizer gags on names with
	spaces for some reason.

2001-03-22 R. Steven Rainwater <steve@ncc.com>

	* article.c (article_form_serve): Removed Post button. User is now
	required to use the Preview button before posting. Added more
	descriptive comments about the use of the posting form (hard-coded
	for robots.net).
	
	* article.c (article_generic_submit_serve): Added more descriptive
	comments about the use of the article preview form (hard-coded for
	robots.net).

2001-03-16 R. Steven Rainwater <steve@ncc.com>

	* article.c (article_render_topic): Added "alt" attributes to topic 
	icons for HTML compliance.

	* site.c (site_render): Fixed bug that was putting illegal </img>
	tags into the XML-generated HTML output. These tags were causing
	rendering problems on some older browsers. Also fixed a number of
	case-sensitivities that were screwing up XML->HTML parsing.

2001-03-15 R. Steven Rainwater <steve@ncc.com>

	* article.c (article_render_from_xml): Article titles on lead stories
	are now links to the story.

	* util.c (render_date): Added optional date render format which
	includes hours, minutes, and time zone information. A new arg was
	added to select the desired render format and render_date() calls
	throughout mod_virgule were modified to use the new calling style.

2001-03-09 R. Steven Rainwater <steve@ncc.com>

        * rss_export.c (rss_render_from_xml): Added some code to strip HTML
	anchor tags from the exported story descriptions. Whiles these 
	appear to be legal in RSS, they are annoying.

2001-03-07 R. Steven Rainwtaer <steve@ncc.com>

	* mod_virgule.c (test_serve): #ifdef'd out this function since it 
	seems to not be needed or useful.

	* proj.c (various): Change hard-coded values used by Advogato into
	hard-code values for robots.net. These really need to be read from
	XML rather than hard-code...

	* site.c (site_serve): If a request for a URI ending in .html cannot
	be filled by an XML document, an attempt is made to find a static
	HTML file before declining the request.

2001-03-02 R. Steven Rainwtaer <steve@ncc.com>

	* acct_maint.c (acct_person_serve): Cert selector is not
	displayed for users when looking at their own page.

	* article.c (all): Changed calls to ok_to_post() in cases where
	ok_to_reply() was more appropriate.

	* article.c (article_form_serve): Changed maximum article title 
	size from 50 to 40 characters.

	* article.c (article_generic_submit_serve): Changed maximum article
	title size from 50 to 40 characters. Fixed bug that caused loss of
	article topic when previewed.
	
	* auth.c (auth_user): Authenticated user name from cookie is set in
	the Apache request_rec->conn_rec->user field so it can be logged.

	* proj.c (proj_newsub_serve,proj_editsub_serve): Now uses the
	ok_to_create_project() function instead of ok_to_post(). Reduced
	project name size from 30 to 25 characters.

	* req.c/h (ok_to_post): User must have journeyer or higher cert to
	post an article.
	
	* req.c/h (ok_to_relpy): New function. User must have apprentice or
	higher cert to post a reply to article.
	
	* req.c/h (ok_to_create_project): New function. User must have 
	apprentice or higher cert to create a project.

	* rss.c (rss_index_serve, rss_render_from_xml): Replaced hard-coded
	references to Advogato with hard-coded references to robots.net.


2001-03-01 R. Steven Rainwater <steve@ncc.com>

	* acct_maint.c (acct_certify_serve): Users are now unable to
	issue meaningless self-certifications.

	* site.c (site_render_banner_ad): Kluge to display some random
	hard-coded banner ads. This will be replaced by real code later.
	
2001-02-27 R. Steven Rainwater <steve@ncc.com>

	* article.c (article_render_from_xml): Slight fixes to article
	header code to achieve a uniform header size and preven wrapping
	in the header.
	
2001-02-26  R. Steven Rainwater <steve@ncc.com>

	* acct_maint.c (acct_index_serve): Removed hard-coded reference
	to Advogato.org from descriptive text on user account page.

	* certs.c (render_cert_level_begin): Removed hard-coded reference
	to advogato. The code is uneeded as the admin user name and cert
	level may be set in the httpd.conf file.

2001-02-25  R. Steven Rainwater <steve@ncc.com>

	* mod_virgule (set_virgule_css): Added this function to read the
	location of the mod_virgule style sheet from the Apache config file.
	
	* mod_virgule (create_virgule_dir_conf, test_page, virgule_handler):
	Added code to support reading the location of the style sheet using
	set_virgule_css().
	
	* req.h (_VirguleReq): Added pointer to CSS location
	
	* site.c (site_render): Removed code that was inserting spurious
	newlines into HTML rendered from the XML files. This fixes many
	of the HTML table rendering problems with older Netscape versions.
	
	* style.c (render_header_raw): Style sheet location loaded from
	Apache config rather than a static one is now inserted into raw
	pages.

2001-02-23  R. Steven Rainwater <steve@ncc.com>

	* util.c, site.c: Merged mod_virgule 1.21 changes. Only additional
	functionality over my existing changes is the use of <project> as
	an alternate to <proj>.

2001-02-22 Raph Levien <raph@acm.org>

	* db_ops.c (add_recent): Uses xmlNewTextChild rather than
	xmlNewChild, so escaping is now correct.
	
	* util.c Added <proj> tag and renderer.
	
	* style.c (render_acceptable_html): Added <proj> to list.
	
	* site.c (site_render_recent_proj): Changed to use 
	render_proj_name (which escapes the URL correctly) rather than
	hand-rolled HTML.
	
	Thanks to R. Steven Rainwater and Matt McClanahan for patches.
	
2001-02-22  R. Steven Rainwater <steve@ncc.com>

	* everything : Updated user name support to allow names with spaces,
	dashes, underscores, and dots. All relevant functions throughout
	mod_virgule were updated to work with the new name format.

2001-02-21  R. Steven Rainwater <steve@ncc.com>

	* acct_maint.c (all): Modified all function using cert levels to 
	work with the changes made in certs.c (see below).

	* article.c (article_render_from_xml): Replaced spacer tags with
	legal HTML alternative. Assorted other HTML fixes. If STYLE is
	defined in style.h, the article title attributes are taken from the
	external CSS.
	
	* article.c (article_render_from_xml): Rewrote code that renders
	article headers. A header table is generated that includes a cell
	for an icon representing the article topic. The styles for the
	topic and author cells are drawn from CSS. All these change take
	effect when STYLE is defined.

	* article.c (article_form_serve): Added topic selector input to
	the article posting form.
	
	* article.c (article_submit_serve): Added support for article topics.
	
	* article.c (article_generic_submit_serve): Added support for article
	topics.
	
	* article.h: added ArticleTopic struct.

	* certs.c/h: Major changes to reconcile an off-by-one error between
	CERT_NONE (was 0) and the lowest cert level (was 1). The 0th cert
	level is now considered the default and used in place of both. This
	simplifies many of the functions, structures, and cert calculations.
	Replaced the hard-coded cert levels created for Advogato with user
	configurable ones that may be set in the httpd.conf file.

	* mod_virgule (command_rec): Added new command records for 
	VirguleCerts, VirguleAdmin, VirguleSeeds, and VirguleTopic. New
	Apache option handler functions were added for each of these.
	
	* mod_virgule (test_page): Now renders seeds, cert levels and admin
	values that were read from httpd.conf.

	* mod_virgule (create_virgule_dir_conf): Seed, cert, admin, and
	topic pointers are now initialized.
	
	* mod_virgule (virgule_handler): Pointers to seed, cert, admin, and 
	topic data is now passed to the VirguleReq struct.
	
	* req.c (all): Various changes to work with new cert level code.
	
	* req.h: Added pointers in _VirguleReq for cert_level array, seed
	array, topic array, and admin name.

	* tmetric.c/.h: Various changes to remove hard coded seed values.
	No changes to trust metric calculation were made.

2001-02-20  R. Steven Rainwater <steve@ncc.com>

	* certs.c (render_cert_level_[begin|end]): If STYLE is defined in 
	style.h, the cert level style is taken from the external CSS rather
	than hard-coded.
	
	* db_ops.c (db_relation_get): Added return value to suppress compiler
	warnings. Not sure why this function exists?
	
	* main_page.c/.h: removed these unused files
	
	* mod_virgule.c (virgule_handler): Removed unecessary reference to
	main_page.c function. Fixed prefix bug that affected the foo.html
	diagnostics page. Added explanatory comments to some functions.
	Removed assorted cruft that appeared to have been used by Raph only
	for initial testing.
	
	* proj.c (proj_index_serv): qsort() is now used to alphabetize the
	project list. An assciated compare function, proj_index_sort() was
	added. Removed unused variables to eliminate compiler warnings.

	* site.c (site_render_recent_acct): Replaced spacer tags with table
	tags for HTML compliance. 
	
	* site.c (site_render_recent_changelog): Replaced spacer tags with
	break tags for HTML compliance.

	* site.c (site_render_recent_proj): Replaced spacer tags with table
	tags for HTML compliance.

	* site.c (site_render): Removed some dead code.  Added #ifdefs to
	exclude calls to render_table_[open|close] when CSS extensions are
	compiled in.
				
	* style.c (render_sitemap): If STYLE is defined, sitemap attributes
	are taken from external CSS rather than hard-coded valueds.
	
	* style.c (render_table_[open|close]): Enclosed in #ifdef to exclude
	these functions when CSS extensions are compiled in.

2001-02-20  R. Steven Rainwater <steve@ncc.com>

	* style.c (render_header_raw): Added support for an external style
	sheet named /mod_virgule.css to make life easier when customizing 
	a mod_virgule site.


2001-01-24  Raph Levien  <raph@acm.org>

	* auth.c (auth_user): Fixes denial of service attack. Thanks
	to R. Steven Rainwater for the patch.

2000-11-07  Raph Levien  <raph@acm.org>

	* req.h:
	* req.c:
	* tmetric.h:
	* tmetric.c: Replaced tmetric.xml with plain-text file, which
	removed a major performance bottleneck.

2000-09-02  Raph Levien  <raph@acm.org>

	* diary.c: Provide separate creation and last update dates.

	* wiki.c: Liberalize InterWiki lexing.

2000-08-24  Raph Levien  <raph@acm.org>

	* wiki.c (wiki_link): Fix embarassing bugs.

	* style.c (render_acceptable_html): Add wiki and person tags.

2000-08-24  Raph Levien  <raph@acm.org>

	* util.h:
	* util.c (nice_htext): Added handlers for special tags (wiki
	and person). First arg is now a VirguleReq rather than a pool.

	* acct_maint.c (acct_person_serve):
	* article.c (article_generic_submit_serve):
	* diary.c (diary_preview_serve, diary_post_serve):
	* proj.c (proj_relsub_serve): Changed first arg of nice_htext
	to vr.

	* req.h: Added render_data table, used by wiki to store
	intermap.

	* wiki.h:
	* wiki.c: Renders InterWiki links.

	* Makefile: Added wiki.o.

Tue Jun  6 22:47:04 2000  Raph Levien  <raph@acm.org>

	* acct_maint.c: Fixed problem of logging into an account
	that is an alias. It now follows the alias link.

	* proj.c (proj_proj_serve): Fixed escaping of project name in
	cases where it includes "+".

	* util.c: Added escape_uri_arg() function that escapes query
	arguments correctly.

2000-04-13  Raph Levien  <raph@gimp.org>

	* util.c (nice_htext): Fixed a brown paper bag bug in which
	nice_htext would run off the end of the buffer.

	* article.c (article_generic_submit_serve): Fixed a bug in which
	the nice_text version of the title was ignored.

	* util.c: Added escape_noniso_char, which translates non-iso
	characters to reasonable replacements. Both nice_text and
	nice_htext use this now. This fixes a vulnerability where some
	browsers (notably Netscape) interpret a tag between 0x8B and 0x9B
	characters as an HTML tag, and at the same time takes care of the
	"dreaded question mark" problem when people post from Windows
	browsers.

Sat Apr  8 13:05:38 2000  Raph Levien  <raph@acm.org>

	* proj.c (proj_proj_serve): Properly escape project name in URI.

	* article.c (article_reply_submit_serve): lead and body are now
	passed to this routine raw so that they don't accumulate <p> tags.

	* util.c: Added "pre" tag to acceptable html.

	* rss_export.c (rss_index_serve): Changed root element to
	lowercase "rss".

	* article.c (article_generic_submit_serve): Check for empty body
	when submitting a reply.

	* article.c (article_render_reply):
	* diary.c (diary_render): Added <a name=""> tags to individual
	replies and diary entries. Thanks to Kelly for the patch.
	
	* diary.c: Fixed segfaulting when diary is empty.

Sat Mar  4 16:14:23 2000  Raph Levien  <raph@acm.org>

	* proj.c:
	* diary.c (diary_post_serve): 
	* acct_maint.c (acct_newsub_serve): Increased add_recent to 50.
	This makes sense now that the xml tags in site.c actually pay
	attention to their arguments.

	* diary.c (validate_key): Patch by R. Steven Rainwater to allow
	key of -1, which autoassigns a new post number. This makes it
	easier to use a client for posting diary entries.

2000-03-04  Laurent Gauthier  <lolo@seul.org>

	* site.c (site_render_recent_changelog): Implemented this function
	to render the <recentlog> tag.

	* site.c (site_render): Added a new tag <recentlog>, that will
	display the most recent diary entries.

	* diary.c (diary_latest_render): New function used to render the
	latest diary update for a given user.

	* diary.h: Added prototype for diary_latest_render().

Thu Mar  2 17:18:30 2000  Raph Levien  <raph@acm.org>

	* diary.c (diary_serve): Fixed prefix bug.

	* acct_maint.c (acct_person_serve): Fixed prefix bug.

	* site.c: Made "recent" tags actually limit the number displayed.

Fri Feb 25 13:41:21 2000  Raph Levien  <raph@acm.org>

	* diary.c (diary_render): Fixed a little bug that got introduced
	with the diary editing code.

Fri Feb 25 12:56:14 2000  Raph Levien  <raph@acm.org>

	* acct_maint.c (acct_person_serve): Made an explicit "[Edit]" link
	for diary editing, rather than just making the date clickable.
	Made the personal information page automatically redirect through
	symlinks. Thus, if the account Foo exists and /person/foo/ is
	requested, it will redirect to /person/Foo/.

Wed Feb 23 23:28:12 2000  Phil Schwan  <pschwan@linuxcare.com>

	* diary.c: Added diary editing.  In order to re-use code already
	present in diary_post_serve and diary_preview_serve, I made them
	get the diary's index from the form field "key".  This key is run
	through validate_key() which does simple bounds checking and returns
	a key suitable for passing to db_xml_put.  Keys are originally set
	in diary_index_serve (adding a new diary entry) and diary_edit_serve.

Wed Feb 23 15:52:20 2000  Raph Levien  <raph@acm.org>

	* acct_maint.c: When you create an account with uppercase
	characters, it now creates an "alias" account with the
	all-lowercase version of the name, pointing to the real account.
	Conversely, it checks the all-lc name on account creation. This
	blocks people from creating accounts colliding in name (all except
	for case) with existing accounts. This was a serious problem
	because ap_table_* is not case sensitive.
	
	* tmetric.c: Changes to work with "alias" profiles.

Thu Feb 17 11:46:34 PST 2000  Manish Singh  <yosh@gimp.org>

	* Makefile: use glib-config and xml-config directly, to avoid
	gnome build time dependancy

	* mod_virgule.c: moved gnome-xml #includes to the beginning, so we
	get xmlNode declared for diary.h

2000-01-17  Raph Levien  <raph@gimp.org>

	* diary.c (diary_export): Initial support for exporting diaries
	as XML files.

	* diary.h: diary_export declaration.

	* acct_maint.c (acct_person_diary_xml_serve): Added
	/person/<username>/diary.xml into namespace served.

Mon Jan 10 11:11:49 2000  Raph Levien  <raph@acm.org>

	* acct_maint.c: Support for "older entries" for diary.

	* diary.h, diary.c: Implemented "older entries" link. Implemented a
	"Preview" button for diary posting.

	* article.h, article.c: Implemented "older articles" link.
	Implemented a "Preview" button for article and reply posting.

	* util.c: Added str_subst utility function, used for previews.

Wed Dec 15 14:51:51 1999  Raph Levien  <raph@acm.org>

	* apache_util.c (read_post_data): bytes_read == 0 is an error
	condition. It was looping infinitely on this.

Wed Dec  8 12:24:49 1999  Raph Levien  <raph@acm.org>

	* db_xml.c (db_xml_free): Fixed a really _dumb_ memory leak.

Fri Nov 26 16:02:35 1999  Raph Levien  <raph@acm.org>

	* acct_maint.c: Shows project relationships, also have graph
	display for playing with dot.

	* article.c: Only display "Reply..." link twice if number of
	replies != 0.

	* db_ops.c: Enforce uniqueness when putting new relations in the
	db.

	* proj.c: Correct url escaping for project names. Also
	sanitychecking of new project names.

	* tmetric.c: Added alan :)

1999-11-15  Raph Levien  <raph@gimp.org>

	* db_ops.c (db_relation_put_field): Naming changes.

	* proj.c: Actually stores the "staff" relationship between project
	and person now.

Mon Nov 15 15:43:47 1999  Raph Levien  <raph@acm.org>

	* proj.c: Project page now works pretty well.

	* schema.c: Another new thing is a schema system for automating
	some of the drudge work in having forms with lots of fields. This
	is used throughout proj.c, and should probably be retrofitted to a
	few other things.

	* db_ops.c: I'm starting on support for relations in the database
	abstraction, which will be useful for multiway relations. For
	example, if person X works on project Y, it will (with one
	operation) store information in both /acct/X and /proj/Y.

	* some other minor fixes too, like alphabetizing the /person/ list
	and so on.
