Release Meeting: 2 June 2015


1. 1.2.0 blocker review

Tasks.Item13378 has been closed and a lot of more work done by Crawford and Jozef. New blockers for 1.2 now are:

2. Proposed Schedule for Alpha / Beta / GA

Next actions

Next meeting - - Monday Jun 15, 2015 1300Z — ReleaseMeeting01x02_20150615


(08:36:26 AM) CDot entered the room.
(08:56:14 AM) Lynnwood  entered the room.
(08:58:46 AM) andreli entered the room.
(09:00:17 AM) gac410: Wohoo good turnout.    Hi everyone    Thanks for accommodating my schedule issues.
(09:01:27 AM) gac410: I think our primary agenda item now is blocker review and decision on when to release  Beta2. 
(09:05:45 AM) gac410: Everyone ready to start?    
(09:05:48 AM) CDot: aye
(09:06:16 AM) jomo entered the room.
(09:06:32 AM) gac410: Okay.   1st,   kudos to CDot and Jozef  ...  for excellent work on the 1.2 utf8 / unicode  update
(09:07:10 AM) CDot: :-)
(09:07:15 AM) gac410: We still have a couple of blockers,      but so far things look good I think.
(09:07:39 AM) gac410: MichaelDaum:    Have you had a chance to give the changes a review?
(09:07:42 AM) ***jomo did nothing -> CDot !!
(09:08:02 AM) MichaelDaum: gac410, nope I am packed with work ... heads down.
(09:08:41 AM) gac410: Has anyone else been working with trunk?  
(09:09:08 AM) CDot: anyone?
(09:09:23 AM) CDot: WikiRingBot: done any work with trunk?
(09:09:23 AM) ***WikiRingBot does not understand done any work with trunk?
(09:09:25 AM) jomo: me - in production ;)
(09:11:21 AM) gac410: Okay... well for blockers.     I take it that one can be deferred  post 1.2?     NFC/NFD issue?
(09:12:13 AM) gac410:    seems to be a 1.1.9 issue,    awaiting feedback from nobody.     Last comment was appears fixed in 1.2
(09:12:27 AM) JulianLevens1 entered the room.
(09:12:47 AM) CDot: jomo: needs to comment on
(09:12:58 AM) gac410: item11262,  Item13139 are both "catchall"   so skipping them.
(09:13:09 AM) jomo: hm.. just testing it in freebsd installation and IMHO here is a bug ... in the natedit the image name is corrupt - but again, i have only os x client...
(09:13:36 AM) jomo:      <img alt='&#268;á&#344;ý.png' height='50' src='%ATTACHURLPATH%/Ä&#140;áÅ&#152;ý.png' width='152' />
(09:13:52 AM) jomo: i got the above in the natedit...
(09:14:06 AM) jomo: wgen switching from wysiwyg
(09:14:10 AM) jomo: when
(09:14:42 AM) CDot: so you edited the topic in wysiwyg, then switched to Nat? OK, testing
(09:15:20 AM) gac410: Jomo,  also if you could comment on  ...  I have not seen any more warnings,  with my perlbrew   Can probably be set to waiting for release.
(09:15:36 AM) jomo: 1.) just saved a new topic 2.) attached the image 3.) edit (default wyiswyg) 4.) switch to natedit...
(09:16:00 AM) jomo: gac410: it is perfect - no more warnings.. ;)
(09:17:00 AM) jomo: conflict - GeorgeClark is editing Item13423. :) :)
(09:17:07 AM) CDot: jomo: ok, I can confirm what you are seeing; there's a double-encoding going on somewhere
(09:17:11 AM) gac410: I just set waiting for release.
(09:17:27 AM) jomo: CDot: huh! ok :) :)
(09:18:40 AM) gac410: MichaelDaum:   You opened   Is that a 1.2. blocker?   It doesn't describe how to recreate it or when/where it occurs.
(09:18:55 AM) ***MichaelDaum clicks
(09:19:02 AM) MichaelDaum: cant remember by item number
(09:19:25 AM) MichaelDaum: well this was easy to repro
(09:19:43 AM) MichaelDaum: I tried to edit or move a table line using editrowplugin
(09:20:15 AM) MichaelDaum: the thing tries to convert an obj to json ... and stumbled upon Foswiki::Attr being a blessed obj yet no TO_JSON method in it
(09:20:20 AM) MichaelDaum: the contained patch fixed it
(09:20:46 AM) CDot: hum... I think a Foswiki::Attr should have a TO_JSON
(09:20:54 AM) MichaelDaum: on 1.1.9?
(09:21:08 AM) CDot: ah, ok. Thought you were on 1.2
(09:21:18 AM) MichaelDaum: those plugins must remain backwards compatible
(09:21:38 AM) MichaelDaum: it should block their new release if thats not the case
(09:22:31 AM) CDot: y. fair enough, your patch is fine then
(09:23:00 AM) gac410: MichaelDaum:   can you apply it then?   
(09:24:00 AM) MichaelDaum: no not now
(09:24:14 AM) MichaelDaum: can somebody please take over?
(09:24:25 AM) gac410: Okay I can apply it
(09:24:29 AM) MichaelDaum: thanks
(09:25:18 AM) gac410:    One more edit row plugin bug I'm uncomfortable with,  but I didn't make it a blocker.   In the TestCaseUtf8Errors page,   one table gets truncated and I have no idea why.
(09:26:39 AM) gac410:   has a "workaround" ... it seems to be a perl bug.   I trap the failure with an eval, and things seem to survive.     
(09:28:05 AM) gac410: Any other 1.2 blocker concerns?  
(09:28:30 AM) CDot: are you using utf8/unicode throughout the i18N/maketext process?
(09:28:31 AM) jomo: gac410 for the 13424 - if you want - test the 2line patch... ;) it helps - but much more encoding occurs...
(09:28:51 AM) CDot: or can people edit/save bytes?
(09:29:08 AM) ***gac410 doesn't understand the question 
(09:29:25 AM) CDot: ok. All files are saved as bytes, right?
(09:29:44 AM) CDot: so, how do people interact with those files that provide the language stuff?
(09:29:55 AM) CDot: is it only through a fully unicode-compliant interface
(09:30:08 AM) gac410: weblate just commits the stuff into github.    I don't manually edit 
(09:30:10 AM) CDot: or can they text-edit the byte files, ignoring the encoding?
(09:30:11 AM) jomo: CDot: it seems to be an perl-error - for some perl versions it works - for some no...
(09:30:36 AM) CDot: if it's a perl error, it really needs nailing down
(09:30:49 AM) ***CDot would supect it's more likely an editing tools error
(09:30:58 AM) CDot: or an error in a CPAN module
(09:30:58 AM) gac410: But it only fails on perl 5.16.3
(09:31:06 AM) CDot: still.
(09:31:22 AM) CDot: until you can nail it down....
(09:31:26 AM) jomo: the same maketext - the 5.16.3 fails - 5.20.2 works - and such...
(09:31:26 AM) gac410: I've tried updating 5.16.3 to very latest Locale / Maketext   extensinos,  still fails. 
(09:32:10 AM) CDot: really need to develop a micro-example and report it upstream
(09:32:21 AM) gac410: is 5.16 still maintained?
(09:32:22 AM) CDot: cos otherwise, such a perl bug could have other repercussions
(09:32:35 AM) gac410: 5.18 works fine
(09:32:53 AM) CDot: hmmm. then don't we have to require 5.18?
(09:33:01 AM) CDot: at least when i18n'ing?
(09:33:36 AM) gac410: With ASSERTS disabled,  the logs are noisy.   It's a warning, not a crash
(09:33:42 AM) CDot: oh, ok
(09:33:48 AM) CDot: I thought it was a crash
(09:33:51 AM) CDot: n.m. then
(09:34:04 AM) gac410: it's actually just the copyright string    and I have no idea why it's just that string in 2-3 languages that crashes.
(09:34:19 AM) gac410: well on trunk it's a crash because we run with asserts  :)
(09:34:24 AM) gac410: t.f.o tha tis.
(09:35:07 AM) jomo: in the net i saw some mentions about the bug in the Encode... in the 5.16...
(09:35:26 AM) gac410: we should probably add in the release notes that there is a know issue with perl 5.16
(09:35:46 AM) CDot: y
(09:36:47 AM) jomo: or just add the 2line patch for the 5.16... - it works (only encodes every maketext output) - not at once at read - as default...
(09:38:14 AM) gac410: But as you say,  we would need to do your patch everywhere i18n->maketext is called.   
(09:38:55 AM) jomo: damm - yes... it is 2 line but many times... youre right... it doen't make sense...
(09:39:49 AM) gac410: So far we've only seen one string that causes the issue on t.f.o,  and the eval I added traps  the error.
(09:40:59 AM) gac410: Damn...   git conflict in t.f.o    >>>>>>> Stashed changes
(09:41:54 AM) gac410: okay.   no big deal.   I fixed LieVen's  "developer question"  which he posted into the Main/Webhome.
(09:43:28 AM) jomo: guys, here are some plans to convert F.O and T.F.O to utf8? (last time i checked the store was iso1 - because the compatibility with the F.O.)
(09:43:36 AM) gac410: And for the impact of Item13424.   View  with language  French,    and copyright string at bottom becomes:   ERROR: Translation failed, see server error log.
(09:43:50 AM) gac410: jomo    That's next on the agenda
(09:44:19 AM) jomo: cool
(09:45:03 AM) gac410: So for blockers.      Just Item13338 - the NatEdit image issue.    and CDot, you are working on it ???
(09:46:00 AM) gac410: And I'll take 13438,   the TO_JSON blessed object issue,     And Item13424 (perl 5.16 error)   I'll document in release notes.
(09:46:31 AM) gac410: Ah...  Unit test crashes.    There are 6 failing testsin the nightly runs
(09:46:44 AM) gac410: I cannot recreate the TWikiFunc failures.  They all work for me,.
(09:47:28 AM) gac410: The RCS errors.    CDot  I looked them over,  and was afraid to 'fix the tests'  as they did seem to be reporting a change in how corrupted TOPICINFO is dealt with.
(09:49:27 AM) gac410: Okay.   I'll take silence as   nothing more to add???    
(09:49:43 AM) gac410: So   onto jomo's question. utf8 change
(09:50:02 AM) gac410: CDot and I started working on that yesterday.   
(09:50:34 AM) gac410: I was going to run the CharsetConverter last night, but held off due to some conversion needed on trunk Sandbox.
(09:51:41 AM) gac410: The converter will only run on 1.1.x installations,   so this evening I plan to symlink the 1.2 webs back into our 1.1.9 install,  shutdown apache,  and run the converter
(09:52:00 AM) CDot: gac410: yes, I'm working on it (hence the lack of attention :-( )
(09:52:06 AM) gac410: :)
(09:52:44 AM) gac410: That brings up a issue we found on f.o.     WikiAppInstallerPlugin     And the    WikiApps   web.
(09:52:47 AM) CDot: the reload tests... I added 'skip' to those tests, but it doesn't seem to be working :-(
(09:53:13 AM) CDot: it works for me locally
(09:53:19 AM) ***CDot checks he pushed it
(09:53:38 AM) gac410: CDot:    The tests fail for me.   Are they reporting a legitimate error?
(09:53:48 AM) CDot: no
(09:54:27 AM) CDot: ah, I didn't push it yet, I was adding some more unit tests
(09:54:41 AM) gac410: i tried changing the asserts to match what was returned,  but it keeps going downhill.   More asserts fail
(09:54:55 AM) gac410: clearly store is returning a different version that what the test wants.
(09:55:33 AM) CDot: y, because the test is *wrong* for 1.2.0, which has the problem fixed
(09:55:59 AM) CDot: it's a bad test.
(09:56:02 AM) gac410: Ah...    okay.
(09:56:02 AM) gac410: So everyone else lurking here.     trunk has some very very major changes that will go into Beta 2 - if you haven't guessed that by now.
(09:56:26 AM) jast: there wasn't _that_ much guesswork involved...
(09:56:30 AM) gac410: L)
(09:56:32 AM) gac410: :)
(09:57:15 AM) gac410: If anyone else can run / test 1.2  before we go public with beta 2,  that would be helpful 
(09:57:34 AM) CDot: especially those with odd clients / strange charsets, please
(09:57:35 AM) jomo: with plainstore it just works... ;)
(09:57:51 AM) jomo: (not counting some minor errors)
(09:58:36 AM) CDot: oh.... and if you have anyone in your teams/clients who have used mixed charsets in the 1.1.9 installs (save UTF-8 in an iso-8859-1 install) then we (gac410 or I) can help you.
(09:58:39 AM) gac410: we probably also need one more review of the UpgradeGuide & release notes.   Do we highlight the changes enough.
(09:58:58 AM) jomo: will try run again the unit-tests (last times whe i tried - it died)...
(09:59:20 AM) CDot: jomo: write a strongly worded email to Apple ;-)
(09:59:54 AM) gac410: Okay so back to      The WikiApp  web  allows guests to update topics via the WikiAppInstallerPlugin
(09:59:57 AM) jomo: heh ;) i'm pretty happy with osx - but this is purely my own mac-filia ;)
(10:00:58 AM) gac410: That one not that it is explained was not as worrysome,   but the issue is that bots  scanning the links on the WikiApp webhome results in updates of sandbox/WikiGuest subweb.
(10:01:44 AM) gac410: So we need decisions on    1)  Should we kill the WikiApps   web and demonstration apps.    They seem to have not been maintained for 4 years.
(10:02:06 AM) CDot: what's there?
(10:02:46 AM) gac410: ContactList,   FamilyTree    and a few others.     co-author is some guy with handle CDot   :D
(10:03:07 AM) CDot: I remember now... we developed it during the first Foswiki camp
(10:03:18 AM) andreli: I would say: kill the web and bring it back later new an shiny
(10:03:18 AM) CDot: Padraig did most of the work, and I helped out a bit
(10:03:51 AM) CDot: it would be a shame to kill it *dead*, but as it is, it's not useful
(10:03:58 AM) CDot: probably only visited by bots
(10:03:59 AM) gac410: Is Padraig still active at all?
(10:04:02 AM) CDot: no
(10:04:11 AM) CDot: once in a while, but not reliable
(10:04:28 AM) CDot: he will respond to email, is about all
(10:04:39 AM) andreli: Ok, not kill it, but put it aside for the time beeing
(10:04:56 AM) gac410: well we could restrict WikiApps to logged in users,  that would get rid of the guest updates, but leave it active for others.
(10:05:09 AM) CDot: y, let's do that
(10:07:00 AM) gac410: Done
(10:07:49 AM) gac410: Okay.  So for tonight.   I will shut down apache.  probably around 0000Z     and do the utf8 migration
(10:08:26 AM) gac410: If I get into trouble,  unless it's a complete disaster,   I'll leave you an email for tomorrow morning if that's okay CDot
(10:08:38 AM) CDot: ok
(10:09:58 AM) gac410: Once we have  trunk.f.o   running with utf8,      give things a few days for new blockers to pop up.
(10:10:10 AM) gac410: Any concerns about building Beta 2   this upcoming weekend?
(10:10:18 AM) jomo: no :)
(10:10:32 AM) jomo: only one - call it as 1.9 :)
(10:10:42 AM) CDot: 2.0.0?
(10:10:55 AM) jomo: nah - it isn't 2.0...
(10:11:09 AM) jomo: the 2.0 should be PSGI clean... ;)
(10:11:33 AM) CDot: true, and have "More topic actions...." ajaxified
(10:11:40 AM) jomo: (imho much more work as with the utf8 switch)...
(10:11:50 AM) gac410: I think it will be less confusing to stick with 1.2.0   It's all we've been talking about for a long time.   I don't think we should skip it.  It would imply something was wrong.
(10:11:57 AM) CDot: k
(10:12:04 AM) jomo: ofc - just kidiing ;)
(10:12:19 AM) gac410: MichaelDaum:  is our marketing savvy guy ..  
(10:13:48 AM) gac410: So for existing sites.      advice is:     bulk_copy .pl  to convert to utf8 ... unless your topics have mixed character sets.   
(10:14:08 AM) gac410: If mixed,   then install CharsetConverterContrib   and  "-r" to repair
(10:14:56 AM) CDot: right
(10:15:02 AM) CDot: before the bulk_copy
(10:15:15 AM) gac410: Oh.    MichaelDaum    One thing I'll do before building 1.2,   is merge the FastCGI and ModPerl   engine contribs into   distro
(10:15:43 AM) gac410: That will be a big repo change,  so everyone, if you have pending commits to the engines,   please get them done 
(10:15:52 AM) CDot: gac410: we should probably check in the local hack to CharSetConverterContrib
(10:16:20 AM) CDot: when are you gooing to do it?
(10:16:20 AM) gac410: CDot:    will the bulk_copy  fail if there are mixed characters?   How does a site know to need CharsetConverter
(10:16:43 AM) CDot: the bulk copy won't fail, it will just transfer garbage
(10:17:00 AM) CDot: since those topics are *already* garbage, that's OK
(10:17:03 AM) gac410: Do it? ...   (merge engin repos? )    I'll give 24 hour notice.  Probably this weekend
(10:17:07 AM) CDot: ok
(10:18:02 AM) gac410: CDot   I did some more hacks to charset converter.     Print the name *before* starting the conversion,  so it's clear which topic triggered errors.
(10:18:10 AM) jast: are you planning to completely rewrite the distro repo for that?
(10:18:19 AM) jast: shouldn't be necessary
(10:18:24 AM) gac410: jast:   No,   it should just be a merge commit
(10:18:38 AM) jast: okay... in that case I don't think it will be that big a change
(10:19:02 AM) gac410: it's more a bigger change for anyone with branches and pending work in the engine contribs themselves.
(10:19:20 AM) jast: that it is, yeah
(10:19:47 AM) jast: still fairly doable, though
(10:20:32 AM) gac410: I documented what I plan to do  ... somewhere  :D
(10:20:57 AM) gac410:
(10:21:03 AM) jast: in any case I'm not worried :)
(10:22:31 AM) gac410: By using the filter-branch,  it avoids the "move" of files, so the log command continues to work without needing the --follow option
(10:23:17 AM) jast: the downside is that all the history is rewritten :}   but I don't actually care that much which way we do it, so there's no need to delay the rest of the meeting on my account
(10:23:53 AM) gac410: Yeah.    it breaks all the links in Tasks.   I'm not sure what to do about that.
(10:24:03 AM) gac410: But they'd be broken even without the filter_branch
(10:24:26 AM) gac410: Lieven has been very active with translations of multiple languages.    
(10:24:41 AM) gac410: nobody has complained ... hopefully it's been acceptable.   
(10:24:48 AM) jast: we could keep an archive copy of the repo(s) somewhere to at least make it possible to find the right commit _somewhere_
(10:25:04 AM) jast: I haven't had much time to review translations
(10:25:18 AM) gac410: I'll leave the repos on github, but with OBSOLETE_   comment
(10:25:19 AM) jast: I'll look at German occasionally, though
(10:25:41 AM) gac410: he's been changing french as well,   I'm hoping RaulFR has been seeing them.
(10:26:24 AM) CDot: do they speak french on Klingonia?
(10:27:07 AM) ***CDot wonders briefly what the home planet of the Klingons is called..... and quickly decides he really doesn't care)
(10:27:28 AM) gac410: Qo'noS
(10:27:46 AM) CDot: I knew I shouldn't have asked
(10:27:58 AM) ***gac410 had to look it up as he doesn't care either.
(10:28:02 AM) CDot: :-)
(10:28:19 AM) jast: so then why did you look it up? ;)
(10:28:48 AM) CDot: Klingonia has a much better ring to it. Onomatopoeic.
(10:28:58 AM) gac410: But regardless Lieven has been very diligent reporting source string errors - missing periods,  extra spaces,  grammar issues, etc.
(10:29:22 AM) gac410: So it's improving our overall quality.  
(10:29:29 AM) CDot: excellent
(10:30:10 AM) gac410: One point he brought up.    "Search"     is used in many places   sometimes as noun,  sometimes as verb.      But in German it should be two different words
(10:30:25 AM) ***jomo wondering than using gettext.js (client-side load of the po.files) could helps performance..
(10:30:42 AM) gac410: or even an adjective    "Search results"    
(10:31:14 AM) jast: I don't think that's an adjective... more like a short form of "results of search"
(10:31:18 AM) CDot: jomo: it does when you have a really fast server, but if your server is a bit slow, it can slow everything down.
(10:31:26 AM) gac410: But for the single word,   does anyone know if maketext has any tricks to  note  usage  as verb or noun?
(10:31:43 AM) jast: well, we still are lacking localization of messages generated on the client side
(10:31:49 AM) jast: (AFAIK)
(10:32:35 AM) jomo: CDot why? it loads the po file only once - and everything is done at browser - what has the server with this? 
(10:33:00 AM) CDot: jomo: oh, sorry, for a single po load then yes, it can help
(10:33:17 AM) ***CDot was working with an AJAX translation lookup, which is a dog
(10:33:40 AM) jomo: ah yes - thats is an overkill
(10:34:10 AM) MichaelDaum: gac410,  no I don't have any changes pending for fcgi
(10:34:45 AM) MichaelDaum: and never been using mod_perl in production envs
(10:35:19 AM) gac410: okay    Then sometime this  weekend, the individual repos will be deprecated.
(10:35:24 AM) MichaelDaum: marketing 1.2. 0.... don't make me writing about KIingon UI :/
(10:35:37 AM) CDot: I have one change for FastCGIEngine, but it's not important and can be skipped (unless jomo wants to pursue NFC/NFD normalisation)
(10:36:22 AM) CDot: MichaelDaum: what have you got against the Klingons? Are you a Romulan sympathiser? ;-)
(10:36:22 AM) jomo: nah - sure not now... problems will shown itself when the FO/TFO will runs 1.2 ;)
(10:37:07 AM) gac410: Yeah,  but can t*  boast that they can be translated to klingon    :D
(10:37:22 AM) CDot: ok, and I have the same change for ModPerlContrib, also not ciritical
(10:37:31 AM) MichaelDaum: CDot, nerd - nerdier - klingon
(10:38:04 AM) gac410: Engine changes can go before or after.     If you are maintaingin them in a branch,  it would probably be easier to get them done first.
(10:38:26 AM) MichaelDaum: though QM in Klingon ... this is almost brill again
(10:38:35 AM) CDot: y, ok. I can't remember the task number. One of jomo's
(10:38:50 AM) jomo: ?
(10:39:01 AM) CDot: NFC/NFD normalisation
(10:39:40 AM) gac410: well regardless MichaelDaum   supporting Lieven in translation to klingon has resulted in a lot of good quality fixes in the underlying source strings.  So it's a win IMO
(10:40:10 AM) MichaelDaum: gac410, def
(10:40:17 AM) jomo: this?
(10:41:24 AM) gac410: Okay.   Any other concerns for the release meeting?   
(10:43:10 AM) andreli: Looking forward to Beta 2. Keep up the excellent work. Thanks a lot guys.
(10:46:17 AM) jomo: 1.2 will be wonderfull - it correctly shows (and allows edit) extremelly strange uni-sequences too, like:    &#9786;&#804;&#771;&#824;&#8413;
(10:47:52 AM) JulianLevens1 left the room.
(10:48:37 AM) CDot: jomo: perfect, thanks
(10:51:30 AM) MichaelDaum: good to know
(10:51:32 AM) MichaelDaum: :)
(10:52:43 AM) jmk0 entered the room.
(11:18:44 AM) andreli left the room (quit: Remote host closed the connection).
(12:44:36 PM) MichaelDaum left the room (quit: ).
(01:05:39 PM) jomo left the room.
(01:20:22 PM) Lynnwood is now known as Lynnwood_
(01:34:21 PM) CDot left the room.
(01:50:23 PM) Lynnwood_ is now known as Lynnwood
(02:25:15 PM) Lynnwood left the room (quit: Ping timeout: 256 seconds).
(05:53:33 PM) gac410: I've updated the perl version guidance.
(05:53:35 PM) gac410: Perl 5.8.8 is the minimum perl version. The newer the better. There are  know issues with international characters and Perl 5.16.x   See Item13424. For international character set support, Perl 5.18+ is recommended. 
Topic revision: r3 - 13 Jun 2015, GeorgeClark - This page was cached on 13 May 2020 - 22:24.

The copyright of the content on this website is held by the contributing authors, except where stated elsewhere. See Copyright Statement. Creative Commons License    Legal Imprint    Privacy Policy