Test post for the libxml2 HTML parser

In theory, the libxml2 HTML parser should make best guesses on how to fix screwed up, illegal HTML and all tags should get closed at the end of this diary entry, preventing problems in diary entries that follow or elsewhere on the page.

bold tag with no close

italics tag with no close

strike tag with no close

Advogato Status Report

My New Year's resolution is to start doing monthly status reports again! Here's the first one.

Even though I haven't posted a status update in a while, minor code updates have continued. To find out what's changed in the live mod_virgule code running Advogato, see the changelog. It's always there and nearly always up to date.

The biggest change has been in the XML file store locking code. The previous system relied on a site-wide read/write lock that locked out access to the entire database when writes were happening. This was getting to be a problem because of trust recalculations and diary syndication that happens at the top of the hour. Write locks were often clogging things up for 10 to 15 minutes per hour.

But it's all good now. All the locking code has been totally ripped out and replaced with file-level locking. There should almost never be any detectable site delays caused by locking now. Besides fixing the hourly slowdowns, this also gives us a little more breathing room to continue growing.

Another recent change is a patch from fzort that improves the HTML parsing code to eliminate undesirable tag attributes. The long-term the plan is still switching to libxml2's HTML parser and junking the one in mod_virgule but, until then, this should make things a little more secure.

A few other fixes and improvements:

The GUID of syndicated blog posts is now preserved when they go out on the Advogato diary RSS feed.

Mod_virgule now has built in support for Google Analytics. Drop your GA ID code into the config.xml and the appropriate GA markup appears on every page throughout the site.

Joe Presbrey of MIT contributed a patch for an external FOAF URI on the user profile. This allows you to link your Advogato FOAF to any other existing FOAF profile you may have, helping to consolidate your online identify.

The computed trust level for each user is now exported via FOAF, referencing a local RDF schema that describes the trust levels. This mechanism was suggested by Sean B. Palmer and Dan Connolly on the W3C #swig IRC channel.

Advogato Status Report

A new rev of mod_virgule code is live on Advogato. See the changelog for the details. Here are a few highlights.

A discussion between ncm, raph, and chrisd speculated on why there seemed to be a decline in Google rankings for individual blog content on Advogato lately. It was suggested that a change in the Google ranking algorithm may be placing less value on pages with dynamic URLs like http://www.advogato.org/person/ncm/diary.html?start=191. Advogato has long had static URLs for individual articles, so I've added similar support for each individual blog post. If you click the permalink marker beside one of your blog posts, you'll see it now goes to a static URL with just that one post on the page instead of to a dynamic URL that includes a range of posts. For example: http://www.advogato.org/person/ncm/diary/190.html. The old, dynamic system is still in place so search engines and existing links will get to the right place, of course. There's another advantage to having the static URLs to individual blog entries. These will be used for comment pages eventually. Yes, blog comments are really coming. I promise. Some day.

There's also a fix to minor foaf:mbox_sha1sum bug that was noticed by Andreas Harth.

You may have noticed that our Italian cittaditorino spammers were back with a vengence the last couple of weeks. The community spam flagging system seems to be controlling them. Most of the bogus accounts are being deleted within a few days of creation. At ncm's suggestion, I've added rel="nofollows" attributes to all links to untrusted users in the recentlog, recent people joining list, and Advogato People index. There were already nofollows on all links created by untrusted users but this new addition should prevent search engines from even indexing their profile and blog pages. With all these spam control measures in place, keep in mind it's a little harder than it used to be for real users to create an Advogato account and get certified. Well-known users aren't having much trouble and the new trust injected by adding mako as a seed has helped tremendously. But there are users here and there who haven't collected enough certs to become trusted, like pabs3.

That's all the news for now but more new features are on the way.

The URL rendering bug that redi spotted has been fixed, I think. Looks like it was an artifact of the Apache APR 1.3 to 2.0 upgrade that had gone unnoticed for a quite a while. If anyone spots any other URL issues in the project section, let me know.

Advogato Status Report

A new rev of mod_virgule code is live on Advogato. See the changelog for the details.

Aside from the usual minor bugfixes and tweaks, there are two new features you may have noticed already.

New certification indicators: A visual indication is now added to trust certifications that are less than 30 days old. This should make it easier to spot new certs on the user profiles. You can check this out on your own user profile if you've certified anyone, or been certified by anyone, in the last 30 days.

Article lists: Ever wonder how many Advogato articles you've posted? Or wanted to read other articles by a particular poster? Each user profile now includes a reverse chronological list of the 10 most recent articles posted by that user. For users who are more prolific, there is a link to a separate page that includes a complete listing of all articles posted by that user.

In addition to providing a new way to explore Advogato's articles, this should provide another direct route for search engine robots to find the static links to the articles.