Meeting minutes 2012 06 05

Attendance: Robert Anderson, Jarno Elovirta, George Bina, Radu Coravu, Eric Sirois; Regrets: Eliot Kimber

Item 1: 1.6 Milestone 5 status

Packages created; doc updates completed yesterday. Just before build, updated to the latest version of Ant from 1.7.1 to 1.8.4, and ICU4J. Ant no longer bundles Xerces, so that is also added (version updated from 2.9 to 2.11). First time we have built without Jason - so need to verify that everything is included. Jason previously built on a local machine rather than building out of Git. Only thing missing from Git right now is a copy of Ant. Packages uploaded, doc updated.

Robert uploaded all remaining internal tests from IBM; no more tests hidden inside the firewall. All include a readme with the purpose of the test, to help with integration into the automated test suite.

Item 2: Milestone 6 plans

Final milestone, so limit any major updates. Need to ensure release notes, migration notes, etc are all up to date. Suggestion: add a section with highlights related to new documentation (error messages).

Jarno: few major plans. Would like to work on issue 1207 - proportional table widths - has existed since 1.5.3. Radu has posted a suggested fix.

Robert: help clean up error messages. If we move some items from the "response" to the prose of the message guide, may also want to add more information from the Java code into the error message. For example: if you don't specify args.input, says that input parameter was wrong. Doesn't say missing, file not found, etc. Should fix that. Based on the DITA-OT usage survey Jarno posted, this is a major item. Note - only got 43 responses, which is small compared to the number of subscribers on dita-users. In Java code - have some common error messages with no detail, used in many areas.

George: what we found very useful is when the error code is used as anchor in the doc, so in this way we can create a URL to link directly to the error. Jarno and Robert both thought this was working today but it is not. Jarno will update.

Radu: right now we try to catch the console messages, and give those to the user in another view.

For those unaware: one of the earlier test builds of 1.6 added a new topic that lists all DITA-OT error messages, usually with new and additional explanations. To find the message topic in the latest "development" copy of the documentation, go to: http://dita-ot.sourceforge.net/dev/ Find under "User guide" -> "Error messages"

George: In future - can tie this meeting in with gotomeeting, have an existing subscription.

Jarno: other items for 1.6: would like to remove deprecated items. The ditatargets.xml file includes some working code (generalization), but also re-implements some code, so cannot remove until the single-use code exists elsewhere. Will remove conductor.xml, which has been deprecated many years.

Item 3: Discussion of upcoming changes

Discussion of PDF refactoring - Jarno has removed the *1.0 stylesheets for PDF, which should really simplify working with some of the PDF code.

Another common user problem is memory usage; Jarno has added a toggle to stop generating @xtrf/@xtrc. Plus side: better memory management, because with large amounts of content it will end up using a lot of DOM space. Down side: harder to track down error messages. George points out that some people have hooked into these to do additional processing; DITA for Publishers makes use of it when working with images. Default is still to generate these values.

General discussion of the WORKDIR PI and related - Jarno updating with goal to get away from hard-coded system filepath. Move to system to make it possible to pass the DOM straight through (eliminate I/O on hard drive), or grab from elsewhere, etc.

Transform types: which get the most attention? See the latest documentation - now includes description of each transform type, along with some indication of status. Generally encounter the most issues with RTF, which was created with a limited scope and has not been expanded much. The lesser-used ODT output generally seems more reliable; may be able to reuse that logic to get to Word XML?

General discussion of documentation, contributor involvement. Many thanks to George and Radu from Oxygen for regularly passing along user-reported defects, often including fixes.