Meeting minutes 2015 07 02

Attendance: George, Radu, Jarno, Robert, Eliot, Guillaume, Eric, Roger, Kris

Item 1: Any updates about prior releases?

DITA-OT 2.1 released last week, officially now counts as the "prior release".

Issue already with line ends in "ant" file in the tar.gz package, so have something for a 2.1.1 patch. Probably wait another week to see if anything else comes in. There is a workaround for the line end issue.

2.0: Eliot found bug in chunk processing, was unable to track down the code issue in the time available. Came up in D4P documents with "chapter" topic as root, and subsection as nested topic - want to chunk subsections out. There is a reason everybody is afraid of the chunk module code. Guessing an off-by-one error, but might be easier to rewrite that code than to debug and fix. Jarno cleaned up the code about a year ago, but didn't change approach - nobody likes the code. Agree that probably easier to re-implement than to try and fix, but Jarno would probably use Java, where Eliot would lean to XSLT. Regardless of language, Jarno suggests, would be easier to split into two steps - find break / combine points, then handle them all - easier to create tests. Or, one step that does all combining, another that does splitting. Guessing at least 2 weeks to implement; don't have good set of unit tests to validate it. Robert explains some background; adds that chunking problems in IBM tend to resolve around linking (link to a subtopic or element in a topic that gets split/combined).

For older releases 2.0, 1.8, we have some fixes outstanding, just haven't built / released packages. Anybody can build from those, but most dev effort now is going into 2.2 / DITA 1.3.

Item 1.5: startcmd

It ships with 2.1 after some last minute havoc in the code. Robert to do PR about it this month, ideally with input / additional help from Roger in the docs. Will consider dropping from 2.2 but no official decision yet.

Item 2: DITA-OT 2.2 plans and status

Well, calling it 2.2 for now, could be 3.0 if it has enough big major new stuff in it.

Travis testing is currently broken - not sure why. Work on Jarno's mac, on Robert's Windows machine, but Travis runs on Linux. Has never been a problem in past, and difficult to debug; if it works on OS-X, it should work on Linux. When Jarno adds something to develop branch, can take 5 or 15 min to get test results, depending on availability of machines at Travis. Hesitant to add other new function until we debug this, because could break something else without realizing.

• URI based processing - has caused the Travis issue - done by converting file objects to URIs. Lets DITA-OT access HTTP as input argument, which is nice and shiny. Code hasn't been particularly optimized, could have network issues reading from source, so probably don't want to use this all the time, but it's a start.
• Moving map-merge up ahead of keyref/conref processing. Code is ready but waiting to address URI issue before checking in. Switching order improves / simplifies processing, still passes integration tests. Working on this also brought out a few small bugs in current key implementation.
• DITA 1.3 implementations - moving ahead, probably work on branch filter after figuring out URI issues.

Integration testing - Eliot set up Windows server for D4P testing, could be used for DITA-OT testing as well. Can't take on setting all that up but can make machine available.

Item 3: Doc updates and plans

In addition to startcmd stuff - probably want to write up use of Properties file, for DITA 2.x. Also: maybe generate docs on available extension points? Depends on what is installed, but could be helpful for default plugins, and make sure we don't miss any. Jarno has code that could generate that, issue is, have to decide what to generate. One list with all extensions, let HTML toggle by plugin? One reference topic for each plugin? Same thing that applies to arguments that a plugin accepts. Maybe generate both, can pick which you want... If urgent, Jarno can hand over code to finish off.

Item 4: DITA-OT Day 2015

Starting to think about session topics. Discussion going in #dita-ot-day-2015 channel on Slack.

George has been looking over docs from last year. Try vendor panel again? Gave up on it last time - probably will plan not to have it, can rethink later if needed. George working up schedule, then can look for sessions to fill. Will be looking for people to submit session ideas. Last year we thought up most major sessions, had others submit ideas to fill out agenda. Not sure if we will do the same or start off asking for submissions. Would like to take the next week to think up sessions and see what we have.

Already have vendors willing to support the conference, so that's great.

Could already start promoting, but George would like to have call for proposals open when we do that. So look for that end of next week - after we spend week coming up with starter ideas for sessions.

Event info here: http://www.oxygenxml.com/events/2015/dita-ot_day.html

Item 5: OASIS DITA 1.3 status

Voted this week to move to public review; should begin next week.

Item 6: Other topics

Next meeting - August 6, same time (10 AM US Eastern).