January 16, 2003


We're not there yet. The latest build of Syndirella is not yet ready to be called version 0.9. But it has some important changes that needed to be done before the release.

First of all, Syndirella now processes all unhandled exceptions and automatically creates e-mail reports to me containing the stack trace of the exception. I think that this way to report exceptions is most convenient both for me and for the users - much better than transmitting exception reports by HTTP. First of all, I don't need to build any system for receiving the reports. Second, I automatically get the e-mail address of the user, so that I can follow up to him on the report. Third, users need not be concerned about their privacy - they will see all the data that is sent to me.

The only problem with that method is that some mailers (most notably, The Bat!, which is commonly used in Russia, and which I use myself) cannot handle mailto: URLs longer than 255 characters, and that is obviously far from enough for a complete stack trace. However, both Outlook and Outlook Express (running under WinXP) have no problems with the URL length. (I'm afraid that Win98 could also limit the URL length. I'll have to check.)

Besides that, I've finally settled the licensing issues. The author of the collapsible splitter control has given permission to distribute it under the GPL, so its source is now included in the Syndirella source archive. However, the UtilityLibrary will be distributed in binary form only, with a special exception to the GPL allowing that.

As for the code changes:

  • Added a button to the About box to show the Syndirella license.

  • Improved the RSS autodiscovery. Now it really implements all steps of Mark Pilgrim's method - it also detects links which contain rss, rdf or xml in the link text, not only in the link end. Also, local links are detected more correctly. Thus, the feed for Ingo Rammer's weblog is now discovered correctly.

  • Rewritten the code for initializing and validating the common feed properties (update interval and old item count). Now you won't get any crashes because of incorrect values for these fields.

  • Added a new option "Unlimited" as the count of old items stored.

The binary and source snapshots are now distributed in separate archives.

Posted by yole at January 16, 2003 01:33 AM | TrackBack
Post a comment