meetings/irc: opensync-20090122.log

10:01 -!- dgollub changed the topic of #opensync to: OpenSync Project IRC Meeting | please no support questions / bug reports ... please wait until the meeting ends - approx. 10:00 UTC
10:01 -!- andieh [n=andieh@wlan0908.rz.uni-leipzig.de] has joined #opensync
10:02 < dgollub> let's tart?
10:02 < dgollub> (tart is also a new word :P)
10:02 < bricks> of course
10:02 < dgollub> who's present?
10:02  * dgollub present
10:03  * bricks present
10:03 < dgollub> (everyone still sleeping)
10:03 < bricks> think so too
10:04 < dgollub> anyway ... let's start with the first topic
10:04 < dgollub> 1. Action Items from previous Meetings
10:04 < dgollub>   * AI dgollub: assemble list of terms. to use for developer API documentation
10:04 < dgollub>                 #1047
10:04 < dgollub>   * AI dgollub: introduce example-plugin using static-capabilities #1048
10:04 < dgollub>   * AI dgollub: document each plugin function - which bits of information can
10:04 < dgollub>                 be set/reported at each point - entry point documentation  
10:04 < dgollub>                 #1049
10:04 < dgollub>   * AI dgollub: doing engine hacking alread, do documentation once hacking is
10:04 < dgollub>                 done #932
10:04 < dgollub>   * AI dgollub: create proposal for new OSyncDB interface #1050
10:04 < dgollub>   * AI dgollub: write initial mail about this topic on opensync-devel@ to
10:04 < dgollub>                 discuss implementation details, and if this is useful at all
10:04 < dgollub> as discussed last time .. i created ticket for most
10:04 < dgollub> expect one or so (which is about writing a mail)
10:04 < bricks> i already saw the new tickets
10:05 < bricks> nice work ;)
10:05 < dgollub> but yeah.. i didn't managed to work on any really hard.. i'm still wasting lots of time on the mixed-up syncing
10:05 < dgollub> but i'm getting closer ... expect slow-sync the mixed-objtype-syncing is working
10:07 < dgollub> #1047 (the API docu terms...) ... could someone else look into this?
10:07 < dgollub> maybe we should use the wiki/terminology page to collect further?
10:07 < bricks> sec let me check the ticket
10:08 < bricks> should be trivial
10:08  * ChrisH partly present
10:08 < dgollub> http://opensync.org/wiki/glossary
10:09 < bricks> ChrisH: could you do that? do you have some free time?
10:09 < dgollub> bricks: i guess the problem is to collect the terms which are widely used
10:09 < bricks> in my opinion most of the terms are already in the wiki
10:09 < dgollub> i would say we just drop this for now... until we do the final API review and finalize the api docuemntation
10:09 < dgollub> which is after 0.39
10:10 < bricks> ok
10:10 < dgollub> and really conenctrate on 0.39 things
10:10 < bricks> i change the version of the ticket to 0.39
10:11 < dgollub> to 0.39?!
10:11 < bricks> is that wrong?
10:11 < dgollub> ok .. version / milestone?
10:11 < bricks> a mixed up version and milestone
10:12 -!- Netsplit wolfe.freenode.net <-> irc.freenode.net quits: lifeless, Brinky, varadhan
10:12 < dgollub> by version you really mean version - right? ;)
10:12 < dgollub> i guess the version thing is about in which version this got found
10:12 -!- Netsplit over, joins: varadhan, lifeless, Brinky
10:12 < bricks> yep think so too
10:12 < bricks> mixed up both terms
10:13 < bricks> btw. I nearly finished http://www.opensync.org/ticket/898
10:13 < dgollub> anyway .. the idea is, to not waste any time for 0.39 on tthis ticket
10:13 < dgollub> the #1047
10:14 < dgollub> and look into this when working toward 0.40-release
10:14 < dgollub> bricks: cool!
10:14 < bricks> the missing stuff is about private functions
10:15 < bricks> they need to be renamened and placed in a _private header
10:15 < dgollub> ok i see
10:16 < dgollub> next topic?
10:16 < bricks> next topic
10:16 < dgollub> (summary: i try to catchup on the AIs... :/)
10:16 < dgollub> 2. User vs. Developer documentation (from (last, last) last meeting)
10:16 < dgollub> is Tuju around?
10:17 < dgollub> if not.. i would suggest to drop this item from the list .. and discuss this on the mailinglist earliest once 0.39 got released not before
10:17 < dgollub> in my  opionen it makes no sense to write any documentation of stuff which is not working at all ;)
10:17 < bricks> :)
10:18 < dgollub> everyone fine with this? next topic?
10:18 < bricks> fine with me
10:18 < bricks> next
10:18 < dgollub> 3. Porting to different platforms * which priority have porting issues?
10:18 < dgollub> i defenitly want to see opensync running on macosx, windows and freebsd, opensolaris and so on
10:18 < bricks> i don't think that proting issues should block 0.40
10:18 < dgollub> ok - right.. this would be my next question
10:19 < dgollub> so i would suggest we have following propsal:
10:19 < bricks> me too but the thread thing could be hard to fix
10:19 < dgollub> we accept all kind of tickets.. but non get assigned to a milestone prior or to 0.40
10:19 < bricks> but i am really happy that opensync compiles on mac at least
10:20 < dgollub> the thread thing i guess would not cause any api change.. it's just cleaning up thread implementation or so .. i really guess we do some wrong kind of thread-ref-couting or so ..
10:20 < dgollub> yeaeh -- it builds, let's ship it :P
10:20 < bricks> :D
10:20 < dgollub> ok next topic
10:20 < dgollub> 4. Testsuite status * unported unittests * disabled testcases 
10:21 < bricks> should we fix all testcases before 0.39?
10:21 < dgollub> at least we could try
10:21 < dgollub> we still can completely drop testcases if they are just wrong
10:21 < bricks> do you know the test coverage?
10:21 < dgollub> yep
10:21 < dgollub> it's approx. 75%
10:22 < bricks> uiii should be higher
10:22 < dgollub> but yeah .. i have written any testcases for to improve the coverage at all.. so it's still pretty good
10:22 < dgollub> i could be worse
10:22 < bricks> of course
10:22 < dgollub> but we can still write more for 0.40 .. once 0.39 got released
10:22 < dgollub> it's quite simple to increase the C0 coverage
10:23 < dgollub> you know that you can get coverage results via CDash and Ctest ?
10:23 < bricks> no i didn't
10:23 < dgollub> there is a build-type called "Profiling" (which is a bit misleading) but this one builds with gcov
10:23 < dgollub> and if you run make Experimental you get coverage summary even on cdash
10:24 < dgollub> http://opensync.org/testing/index.php?project=OpenSync&date=2009-01-22
10:24 < bricks> another cool feature of cmake/ctest
10:24 < bricks> wow thats great
10:24 < dgollub> check "Coverage" .. ther is a host "rincewind" which build with profiling or at least some cflags which also use gcov
10:24 < dgollub> you can also get results for the file.. and the number of calls per line
10:25 < dgollub> arg ..  ok .. the reason why the coverage is quite low is tdue to one test case which is nearly completely not ported
10:25 < dgollub> ./tests/sync-tests/check_sync_error.c   1.92%  1432
10:25 < dgollub> 1432 lines not covered
10:26 < ChrisH> sorry.... customer system crashed, Need to assist there.
10:27 < bricks> next steps for testcases: get all test cases fixed before 0.39 and extend the coverage for 0.40
10:27 < dgollub> yep
10:27 < bricks> fine
10:27 < bricks> next?
10:27 < dgollub> btw. for the other platforms we could also do tests conditional in cmake
10:28 < dgollub> so they get disabled on platforms which are known to have porting issues
10:28 < dgollub> not quite sure if it worth it
10:29 < dgollub> and solaris seem to fail approx. 39 testcases
10:29 < dgollub> should we disable them additioanly conditionally on the platform? so we can really track here regressions as well?
10:29 -!- nordi [n=nordi@p549A2515.dip0.t-ipconnect.de] has joined #opensync
10:29 < bricks> i am not sure
10:30 < dgollub> so we get really always stable results with 0 failing testcases... like we do with "aachen"
10:30 < dgollub> because right now .. i don't really look at all the results of the other platforms.. they're way to maining failing testcases at once
10:30 < bricks> yeah maybe to stabilize the port and the activate the missing testcases step by step 
10:31 < dgollub> IF ( APPLE )
10:31 < dgollub> OSYNC_TESTCASE_DISABLED( engine engine_init "APPLETICKET" )
10:31 < dgollub> ELSE ( APPLE )
10:31 < dgollub> OSYNC_TESTCASE( engine engine_init )
10:31 < dgollub> ENDIF ( APPLE )
10:31 < dgollub> something like that
10:31 < bricks> create tickets for failling testcases on a plattform and the disable them via cmake?
10:32 < bricks> yes i can do that
10:32 < dgollub> yeah .. what do you think? is it worth? .. we could also do "collection tickets" for entire untis of platforms
10:32 < bricks> should be really easy for solaris or freebsd too
10:32 < dgollub> since i'm quite sure most are really based on one single porting bugs
10:32 < bricks> if we track the failing testcases it should be fine
10:32 < bricks> i guess so too!
10:32 < dgollub> how many are failing on macosx appro.x?
10:32 < dgollub> more then 100?
10:33 < bricks> 80?
10:33 < dgollub> http://opensync.org/testing/buildSummary.php?buildid=4399
10:33 < bricks> no 28
10:33 < bricks> wow that's pretty good!
10:33 < bricks> :D
10:34 < dgollub> ok .. it seems to be fail acroos 5 different unittests
10:34 < dgollub> just create 5 tickets instead of 28
10:34 < dgollub> http://opensync.org/testing/viewTest.php?onlyfailed&buildid=4399 
10:34 < dgollub> the symbols one is a scripting thing ;)
10:35 < bricks> ;)
10:35 < dgollub> and feel free to chang the order in the build environemnt
10:35 < dgollub> so you can have one conditional block of disabled testcases
10:36 < bricks> ok
10:36 < dgollub> but only disable them in CMAkeLists.txt ;) not in the code
10:36 < dgollub> ok cool
10:36 < bricks> of course
10:36 < bricks> your code in cmake is really greate
10:36 < bricks> i like that very much
10:36 < dgollub> i ask Samm later if he can do this for freebsd
10:36 < bricks> great
10:37 < dgollub> * unported unittests
10:37 < bricks> and mabye ChisH can do this fpr solaris
10:37 < dgollub> yeah .. good point - i'll ask him later
10:37 < bricks> i guess that mac os and freebsd problems are similar
10:37 < dgollub> regarding uported unittests - there are still some unported units ... it's not that complex task .. maybe someone could look into that
10:37 < bricks> and related
10:37 < dgollub> (i put this in the minutes .. so maybe someone picks this up)
10:38 < bricks> :)
10:38 < dgollub> i guess i can even do a trivial ticket out of this
10:38 < dgollub> since it's bascially always the same change on the code over and over again
10:38 < dgollub> AI dgollub: create TRIVIAL-tickets for unported testcases
10:38 < dgollub> ok next
10:38 < dgollub> 5. OpenSync 0.39 release * what needs to be done to get 0.39 out?  * file tickets and assign to 0.39 milestone 
10:39 < bricks> thank's for this topic
10:39 < dgollub> ok .. i have 3 things so far in mind which have (for me) highest priority:
10:39 < bricks> i would have added it too
10:39 < dgollub> - mixed-objtype-sinking
10:39 < dgollub> - the static/dynamic config thing requested by henrik
10:39 < dgollub> - format-independent capabilities interface
10:39 < bricks> reminder bricks: read the ticket about mixed objtype
10:39 < bricks> :)
10:40 < dgollub> (- API clean up ... break hashtable  api/ osync db api/ ...)
10:40 < bricks> uhh the capability thing
10:40 < bricks> it's like the sword of Damocles
10:40 < dgollub> not quite sure yet if this is really that big "external interface" change.. 
10:41 < dgollub> but yeah .. i still need to write a mail about this on opensync-devel@ how this could look like
10:41 < dgollub> is there anything else which i'm missing
10:41 < dgollub> i really try to focus the next days/weeks really on this
10:41 < bricks> looked at the roadmap some mins ago and i don't see anything else
10:41 < bricks> mssing
10:42 < dgollub> once this is done.. the testcases get reviewed/fixed
10:42 < dgollub> i hope there is nothing file again 0.40 milestone.. there is still screening of 0.40 required.. there are tons of non-OpenSync related tickets failed against 0.40 ..  lots of plugin things and so
10:43 < dgollub> i guess i'll write a script and unassign everything from "OpenSync 0.40" which is not starting with OpenSync as component
10:43 < dgollub> or  query or so
10:43 < bricks> do you think that most of the api is stable?
10:43 < bricks> or is still something missing
10:44 < dgollub> oh .. this reminds of: the  mapping-summary interface henrik requested
10:44 < dgollub> yeah .. there is some minor cleanup
10:44 < bricks> right that really old ticket
10:44 < dgollub> like  osync_XXX_{nth,num}_YYY() which is using int instead unsigned int
10:44 < bricks> could you assign the tivial changes to me?
10:45 < dgollub> it'S a api change.. but not really hard to port ;) it's jsut recompile in most cases
10:45 < bricks> i can fix it asap
10:46 < dgollub> just looking for the ticket
10:46 < bricks> me too
10:46 < dgollub> #973
10:46 < dgollub> and #974
10:47 < dgollub> and #975 is also kind of related
10:47 < dgollub> acutally we should move all [API] tickets to 0.39 milestone
10:47 < bricks> yep
10:47 < dgollub> and slo all the "Public API 0.40 review" tickets
10:48 < bricks> what about #978 is that fixed already?
10:48 < dgollub> bricks: tip ... add to your CFLAGS "-Wextra" .. you'll get lots of signed and unsigned comparsion warnings
10:48 < dgollub> bricks: regarding the int and unsigend thing... you'll kill tons of thos warning once we cleandup the int / unsigned int mess in our interfaces for the lists (_nth_/_num_)
10:48 < bricks> dgollub: thanks i'll do that
10:50 < dgollub> yeah ... this OSyncXMLField .. need to check - i changed some stuff recently
10:50 < dgollub> when intorducing the mutli-level thing
10:52 -!- ianmartin [n=me@89.241.65.202] has joined #opensync
10:52 < dgollub> hi ianmartin 
10:52 < ianmartin> hi
10:52 < dgollub> ianmartin: perfect timing.. we're just talking about #978
10:53 < dgollub> i guess we can close this one as fixed
10:53 -!- bellmich [n=bellmich@bellus.cms.hu-berlin.de] has joined #opensync
10:54 < ianmartin> yeah
10:55 < dgollub> ok cool - i'll close it
10:56 < dgollub> jfyi, we're right now at topic #5
10:56 < dgollub> 5. OpenSync 0.39 release * what needs to be done to get 0.39 out?  * file tickets and assign to 0.39 milestone 
10:56 < ianmartin> thx
10:56 < dgollub> 10:39 < dgollub> - mixed-objtype-sinking
10:56 < dgollub> 10:39 < dgollub> - the static/dynamic config thing requested by henrik
10:56 < dgollub> 10:39 < dgollub> - format-independent capabilities interface
10:56 < dgollub> -sinking +syncing
10:56 < dgollub> 10:40 < dgollub> (- API clean up ... break hashtable  api/ osync db api/ ...)
10:57 < dgollub> and do the API related tickets  #973, #974 and #975
10:57 < dgollub> bricks: could you assign one of the ticket you want to work on?
10:57 < dgollub> (if you not already have done so)
10:57 < bricks> all tickets are reassigned already ;)
10:58 < dgollub> hehe
10:58 < dgollub> ok ... anything else?
10:59 < dgollub> ok ... then let's close in time
10:59 < dgollub> 6. Next IRC Meeting 
11:00 < bricks> it's in the evening right?
11:00 < dgollub> Date: 2009-01-28T19:00:00Z (Wednesday)
11:00 < dgollub> bricks: yeah
11:00 < dgollub> 20:00 localtime
11:00 < bricks> can't promise to attend
11:00 -!- Cwiiis [n=cwiiis@158.43.2.102] has joined #opensync
11:01 < dgollub> (you know.. so that people from the wildwild west and even more wilder east can join ;)
11:02 < dgollub> currently some of my colleague are in autstralia ...  so we have to deal with 5 different timezones... not funny for those who dial in at 3am :) (they don't have to .. but they still do)
11:02 < bricks> :)
11:03 < dgollub> ok - who takes the minutes next time?
11:03 < dgollub> maybe we should ask who takes the meeting for next "morning"-meeting?
11:04 < dgollub> or who talks the minutes in 2-weeks
11:04  * bricks can do it
11:04 < bricks> ;)
11:04 < dgollub> which one? next week .. or in 2 weeks?
11:04 < bricks> 2 weeks
11:04 < dgollub> ok ... anyone volunteers for being backup?
11:07 < bricks> no big choice for backup :(
11:07 < dgollub> ok - backup: dgollub
11:07 < dgollub> meeting closed