[Libreoffice] minutes of tech. steering call ...

[Michael Meeks]

* Present:
	+ Norbert, Eike, David, Stephan, Michael, Bjoern, Andras,
	  Petr, Caolan, Thorsten, Cedric, Fridrich, Tor, Christian

* Completed Action Items
	+ rotate top #10 easy-hacks on front-page (Bjoern)
	+ poke UI / Translation issue bug#42388 (Andras)
		+ a normal l10n bug...
	+ think about Linux build / up-load work (Caolan)
		+ RedHat to handle it
AA:		+ send ssh keys for access to Fridrich (Caolan)
	+ update wiki to mark 3.3.5 as skipped (Petr)
	+ proposal for closer in 3.4.5 release (Petr)
	+ connect to Fridrich's machine & fix Win32 build failure (Michael, Bjoern)
		+ finally it builds !

* Pending Action Items
        + default to TM safe (non-TDF) branding (Thorsten)
        + enable on-line updates for QA in cross-compiled dailies [in progress] (Kendy)
	+ come up with a list of QA heros for next meeting [in progress] (Rainer)
	+ ask wrt. hacking php to improve easy-hacks presentation (Bjoern)
	+ extract 64bit build hardware from firewall (Kendy)
	+ ask Christian wrt. Mac / PPC (Fridrich)
	+ add unexpected mingw regressions to MingW 'most-annoying' (Rainer)

* Meeting time:
	+ please update the doodle poll:
	  http://www.doodle.com/d7cxxvfhergqhhqg

* Release Engineering update (Petr)
	+ Proposal for a one-off 3.4.5 release
		+ due-to extended 3.5 dev cycle
		+ http://wiki.documentfoundation.org/ReleasePlan/3.5#3.5.0_release
		+ have a Beta0 - 1 week earlier than feature-freeze (just for Cor ;-)
		+ skip Beta 2 - for 3.4.5 RC1
		+ skip Beta 4 - for 3.4.5 RC2/release
		+ does it affect a 3.5.6 release.
	+ Dec 5th feature freeze details (Petr/Michael)
		+ branch immediately on feature-freeze (Petr, Caolan)
			+ new onegit makes this much easier
			+ should we wait for green tinderboxen ? -
			  prolly not, stabilise on branches
		+ double code review not required until nearer RC1
			=> bug fix committer should cherry-pick themselves
		+ tripple cross-company code review for features until nearer RC1
			+ exceptions: artwork, unit-tests until RC1
		+ tinderboxes: will attack the libreoffice-3-5 branch primarily

* QA update (Rainer)
	+ we missed Rainer

* FOSDEM / conference pieces
	+ please submit your LibreOffice dev-room talks to:
	  http://wiki.documentfoundation.org/Marketing/Events/Fosdem2012
	+ block out your 4-5 Feb

* debug-levels / modes clarity (Bjoern/Stephan)
	+ writer debugging, DBG_UTIL symbols all gone to OSL_DEBUG_LEVEL > 2
		+ other things depend on this, so can't build in a product version (?)
	+ missing common understanding of how to use:
		+ NDEBUG, DBG_UTIL, OSL_DEBUG_LEVELs etc.
		+ when are they binary compatible / incompatible etc.
	+ Stephan's master plan
		+ assertions should abort (one day)
		+ all these issues are related.
		+ prefer to move it to the mailing list
		+ gather use-cases there ...
	+ iterative way to get there required (Thorsten)
	+ need to document what is there now (Bjoern)
AA:	+ produce a fleshed out concrete proposal for the public list (Stephan)
	+ assertion failures ?
		+ should they be enabled in product build ?
		+ if ability to recover - why assert ?
		+ tracing - printing progress as we go along
		+ need to capture semantics of tracing, warning, errors etc.
		+ assert == logically impossible, or shouldn't happen ...
	+ bin compat should be orthogonal to asserts/tracing (Norbert)
	+ corrupt input currently being flagged by 'asserts'
	+ problem - old asserts, did not abort => we got lots of them.
	+ DBG_UTIL doesn't work on windows - runtime problems (Tor)
AA:	+ add easy hack to kill PRODUCT in favour of DBG_UTIL (Bjoern)
	+ ideally, done as a patch
		+ needs windows experimentation etc.
		+ where does 'assert' output on windows go ?
	+ what about the 'debugwindow' (ctrl-alt-shift-d)
		+ should we dump it ... yes.

* Make binfilter depend on tail_build
	+ not a big issue, will happen as/when needed or someone wants that.
	+ other modules prevent that currently; cf. 'extensions'
AA:	+ reply to Peter's RFC (Norbert)

* UNO API tagging - compile time flags in .hdl ?
	+ difficulty between uses & impls.
	+ source code, or object level ?
	+ unclear how we can do better guesses ...
		+ in-house extensions etc.
		+ theoretically impossible to derive all uses
		  via. reflection etc.

-- 
michael.meeks at suse.com  <><, Pseudo Engineer, itinerant idiot