Latest revision as of 10:49, 16 December 2009

minutes

• meeting started with Nakata Maho giving an overview of what Japanese community does for release testing. Main Points are:
  • test with TCM, VCLTesttool and do some "free" sanity testing in parallel
  • none of the test areas needs to be complete (but at least some tests are done in each area)
  • a QA manager coordinates and communicates the single tasks
  • communication is done via mailing list with support information at the wiki
  • if builds passed the tests these are approved and released. If there have been no tests at al for a specific build, this will not be released (but stay as RC)

• French and German teams have a similar process (all share the idea to use the wiki to give more information to testers. see JA/QA/Release/template en, DE/3.0.1 Release Test or Fr.openoffice.org/QARC as example.

• Rail suggested to generate screenshots with VCLTesttool for l10n testing. This would help to spot errors in translations but still needs time - and is normally not release critical. If one l10n team likes to get screenshots, you may ask at dev@qa or dev@l10n for help. Thalion72 ( André Schnabel) offered to help with that.

• Some discussion about the problem that for some platforms l10n teams may not find testers. RQ asked to release such builds even without testing.
• common agreement after some discussion:
  • a build that has not at least been successfully installed by a l10n tester should not be released
  • instead l10n teams should be able to announce untested builds (Release Candidates) as long as these builds are clearly named as such (untested)
  • l10n teams are encouraged to ask user to test such builds and provide feedback, so that more builds can be officially released
  • RQ will ask at the distribution project about options to distribute untested versions via regular mirrors

log

(18:05:51) khirano: Hi, Rafaella, Rail and all :)

(18:06:27) khirano: Hi, Maho

(18:06:43) _nakatamaho_: hi

(18:07:33) rafaella: I think that both _nakatamaho_ and khirano wanted to talk about the JA testing model, right?

(18:07:49) _nakatamaho_: yes.

(18:07:52) rafaella: Shall we start with this topic, then?

(18:07:59) _nakatamaho_: yes

(18:08:21) rafaella: Perfect! Please go ahead ... I am very curious!

(18:08:31) # Benutzer tora-japan ist dem Kanal #ooonlc beigetreten

(18:08:45) _nakatamaho_: We have a wiki for release QA

(18:08:59) _nakatamaho_: and here is a translation by kubota-san and me http://wiki.services.openoffice.org/wiki/JA/QA/Release/template_en

(18:09:45) _nakatamaho_: Please refer this wiki

(18:10:16) rafaella: Yes, it's open

(18:10:35) _nakatamaho_: Following

(18:10:43) _nakatamaho_: Please refer the time table.

(18:11:18) _nakatamaho_: then we have what we should do

(18:12:04) _nakatamaho_: First,

(18:12:17) _nakatamaho_: we choose the QA manager,

(18:12:25) _nakatamaho_: Then we create a wiki

(18:12:57) _nakatamaho_: Notify release stopper issue to community

(18:13:49) _nakatamaho_: Notify which VCL TestTool we should use,

(18:13:58) _nakatamaho_: Notify RC builds

(18:14:14) _nakatamaho_: Submit builds to the QATrack,

(18:14:30) _nakatamaho_: Invite QA participants

(18:14:49) _nakatamaho_: (optionally) have QA IRC meeting

(18:14:58) _nakatamaho_: Then start QA.

(18:15:16) _nakatamaho_: We have three ways to test OOo,

(18:15:33) _nakatamaho_: 1. sanity release test in TCM,

(18:15:47) _nakatamaho_: 2. sanity release test with VCLTestTool

(18:15:51) _nakatamaho_: 3. Sanity test

(18:16:46) _nakatamaho_: Testers can choose one of these

(18:16:51) _nakatamaho_: and do tests.

(18:17:33) Rail: _nakatamaho_: what's the difference between 1. and 3.?

(18:17:35) _nakatamaho_: The QA manager approves when at least something is done for the rc builds.

(18:17:51) _nakatamaho_: Moreover,

(18:18:13) _nakatamaho_: If RCn has passed the test (n>=1)

(18:18:28) _nakatamaho_: then we also regard as tested for RC(n+1)

(18:19:04) _nakatamaho_: Usually test will take one week or so,

(18:19:42) _nakatamaho_: We wait until several builds to be approved.

(18:19:55) _nakatamaho_: Then, we raise an issue for distribution

(18:20:22) _nakatamaho_: wait until the bouncer link works,

(18:20:39) _nakatamaho_: Then we do announcement

(18:21:00) _nakatamaho_: how we announce: we write an e-mail to [ja-announce]

(18:21:14) _nakatamaho_: announce@ja.openoffice.org mailing list,

(18:21:37) _nakatamaho_: Then update download page, top page

(18:21:47) _nakatamaho_: Check update page,

(18:21:56) _nakatamaho_: add history section at ja.oo.o,

(18:22:23) _nakatamaho_: and finally, create next wiki for the next release

(18:22:39) _nakatamaho_: That's all.

(18:23:21) khirano: Rail: please see http://wiki.services.openoffice.org/wiki/Sanity_Check_Of_L10n_Builds

(18:23:42) _nakatamaho_: What we test covers a very small parts.

(18:24:26) rafaella: _nakatamaho_, thank you! How many contributors follow this process?

(18:24:44) _nakatamaho_: Usually 5-6. Depends on the release.

(18:25:05) _nakatamaho_: more for 3.0.0 less for 2.4.2

(18:25:29) rafaella: _nakatamaho_, would you recommend this process to the other teams?

(18:25:42) _nakatamaho_: Yes.

(18:26:01) _nakatamaho_: We define what actually do for the release.

(18:26:10) _nakatamaho_: and - do not define what we test.

(18:26:17) _nakatamaho_: sorry

(18:26:31) _nakatamaho_: what I want to say was

(18:26:48) _nakatamaho_: we do really sanity test.

(18:26:55) _nakatamaho_: I think - it doesn't matter.

(18:27:15) # Benutzer sophi ist dem Kanal #ooonlc beigetreten

(18:27:25) _nakatamaho_: I believe

(18:28:03) _nakatamaho_: we will find good testers.

(18:28:03) rafaella: _nakatamaho_, I see - random testing...

(18:28:31) _nakatamaho_: then they will define good test cases

(18:29:40) rafaella: _nakatamaho_, so the focus is really sanity testing, right?

(18:30:00) _nakatamaho_: yes.

(18:30:15) _nakatamaho_: otherwise we cannot approve :)

(18:30:26) _nakatamaho_: the builds ;)

(18:30:31) _nakatamaho_: harder test won't make sense

(18:32:19) rafaella: I would be interested to hear from the other teams, if they are following a similar process

(18:33:54) _nakatamaho_: Rail: how russian builds are released?

(18:34:04) Thalion72: we do it similar in the german project ;)

(18:34:33) rafaella: Thalion72, do you also set up a wiki for each release?

(18:35:24) Thalion72: rafaella: yes

(18:35:34) _nakatamaho_: Great.

(18:35:35) Thalion72: it's the easiest way to get people informed

(18:36:14) khirano: Thalion72: can you give us URL of your QA wiki page?

(18:36:49) Thalion72: hmm .. need to look (Mechtilde is normally providing the pages ;) )

(18:37:10) Rail: _nakatamaho_: qatrack->tcm->IZ

(18:37:15) Thalion72: http://wiki.services.openoffice.org/wiki/DE/3.0.1_Release_Test

(18:37:54) _nakatamaho_: Rail : thanks

(18:38:34) Thalion72: we just give download URL and some basic information at the wiki. We track status of TCM tests (although we use TCM only in some cases)

(18:39:02) _nakatamaho_: Thalion72: do you do other tests?

(18:39:09) Thalion72: if builds are ready for testing, we send an "invitation" to our mailing lists (dev@de, qa@de, users@de)

(18:39:29) _nakatamaho_: same :)

(18:39:29) Thalion72: we do automated testing (normally on 3 platforms)

(18:39:49) sophi: hi all, FR team do the same: track is on TCM, our wiki page is here http://wiki.services.openoffice.org/wiki/Fr.openoffice.org/QARC

(18:39:55) Thalion72: and we ask people to "play around and work" with the RC .. and give feedback via mailing lists

(18:40:14) _nakatamaho_: Okay. Do the testers are (personal) volunteers?

(18:40:32) Thalion72: yes

(18:40:35) sophi: _nakatamaho_: for FR team, yes

(18:41:10) Thalion72: actually I have one offer from a more professional testing team ... but was not yet able to anser this

(18:41:47) _nakatamaho_: I've also asking a professional team...

(18:41:50) Thalion72: I'ts a small company providing services for OOo who need to test new releases for their customers anyway .. so they asked how they could help

(18:42:11) _nakatamaho_: hm

(18:43:26) _nakatamaho_: Rail: in your team, do you have a paritipant from OOo supporting company?

(18:44:49) Rail: we have volunteers

(18:45:10) _nakatamaho_: All: if we prepare "QA" machines that do provide automated results for every possible build,

(18:45:14) Rail: the same applies for tr team

(18:45:27) _nakatamaho_: do you think you want to take as a release criterion?

(18:47:22) Rail: _nakatamaho_: it will be great if we can produce screenshots generated by VCL testtool for all requested locale. What do you think?

(18:47:48) Rail: (of course, before the L10N deadline)

(18:48:41) rafaella: Rail, at least a couple of days before the l10n deadline for l10n fixes....

(18:48:50) Rail: yup

(18:51:04) Thalion72: Rail: would be great .. but would take a lot of time

(18:51:43) Thalion72: in general I'd suggest this to l10n teams who provide a full translation the first time

(18:51:56) Thalion72: normally you do not need the screenshots for each release

(18:52:52) Thalion72: if you "just" need to review new strings in the UI it is easier to use the Application instead of looking at the screenshots (at least this is my experience)

(18:54:23) Rail: in ideal situation we need screenshots only for minor releases

(18:55:08) Rail: maybe twice per release (before and after possible fixes)

(18:56:32) rafaella: Rail, just consider that the screenshots do not capture error messages ... it's only dialogs - if I remember well

(18:57:14) Rail: yes

(18:57:33) # Benutzer MechtiIde ist dem Kanal #ooonlc beigetreten

(18:57:41) Thalion72: Rail: and you still need to go through the screenshots (~1500) .. sometimes it is not easy to identify what dialog you are actually looking at ;)

(18:57:59) _nakatamaho_: Hi Mechtilde! nice to meet you

(18:58:23) MechtiIde: hello

(18:58:51) Thalion72: Rail: anyway ... do you need help with screenshots for 3.1?

(18:58:52) khirano: Hi MechtiIde

(19:00:19) Rail: Thalion72: I produce them myself, but if we speak regarding the whole project...

(19:00:26) # Benutzer RQ ist dem Kanal #ooonlc beigetreten

(19:00:40) Rail: MechtiIde: hey

(19:00:58) Thalion72: Rail: hmm .. as said - I would not suggest this to do screenshots for the whole project on a regular base

(19:01:35) Thalion72: anyway - we are going to be off-topic .. screenshots are not release critical (imo)

(19:02:14) rafaella: Thalion72, Rail but we could encourage people who need such screenshots to request them for their languages ...

(19:02:30) Thalion72: rafaella: yes!

(19:03:18) khirano: Hi RQ

(19:03:21) rafaella: Thalion72, so what is release critical and where do you need more support?

(19:03:21) Thalion72: I could help creating the screenshots - no problem.. but it will take some days (and I need a localized build)

(19:03:26) RQ: hi all

(19:03:39) RQ: am i late?

(19:03:50) rafaella: Hi RQ not sure I met you before....

(19:04:00) RQ: you have, I'm Rimas

(19:04:10) _nakatamaho_: Rafaella: how we can increase the number of release packages?

(19:04:13) _nakatamaho_: Hi RQ

(19:04:14) RQ: the Lithuanian testing lead

(19:04:25) rafaella: Rimas, nice to see you here!

(19:04:29) Thalion72: rafaella: in general I have no trouble with releasing a localized build .. but others may have

(19:04:31) Rail: RQ, date --utc ;)

(19:04:50) RQ: so I am :)

(19:05:10) RQ: anyway, I have one idea that I didn't express last time

(19:05:12) Thalion72: rafaella: we should publich the simlified process .. and the suggestions made by maho at the l10n/qa site

(19:05:37) RQ: the idea is that perhaps we don't have to test on all platforms to release on all?

(19:05:58) RQ: e.g. it's hard to find Linux/PPC or even OSX/PPC testers nowadays

(19:06:10) _nakatamaho_: right

(19:06:18) Rail: and users ;)

(19:06:20) RQ: but I wouldn't want to exclude those builds just because of a lack of testers

(19:06:21) Thalion72: does not like this idea

(19:06:23) rafaella: _nakatamaho_, I think that we need to make it very clear and very simple to all how to release and distribute l10n builds

(19:06:41) _nakatamaho_: very clear

(19:06:49) RQ: Thalion72, you don't?

(19:06:53) MechtiIde: so l10n teams only release for the plattforms they tested

(19:07:05) MechtiIde: not more

(19:07:06) Thalion72: RQ: no .. users should be aware that they use an untested build

(19:07:13) RQ: Thalion72, maybe

(19:07:20) # Benutzer thorstenziehm_ ist dem Kanal #ooonlc beigetreten

(19:07:30) _nakatamaho_: Hi thostenziehm

(19:07:55) RQ: but OTOH, if a build works on say Linux/x86 and OSX/x86, then even if it breaks on PPC, it's most probably not a localization bug, is it?

(19:08:17) Thalion72: RQ: depents indedd ... e.g. for Germany I would never risk to "release" something that could not be installed at all

(19:08:27) _nakatamaho_: no test, no release. this is simple and clear

(19:08:53) RQ: Thalion72, well, I think if a Lithuanian OSX/ppc build would fail, than most likely any other localized build would too

(19:08:58) Thalion72: RQ: "probably" is the point here .. "probably" not .. but you cannot guarantee

(19:09:09) rafaella: RQ, in principle that's correct and in fact in this case we could set up a global matrix to distrubute the testing work across langauges and platforms

(19:09:20) RQ: exactly

(19:09:24) RQ: OR

(19:09:34) Thalion72: RQ: no .. we had this with polish builds for 3.0.0 .. all localized builds workedfine - only Polish windows builds not

(19:09:38) RQ: some testers could test all builds whether they run at all

(19:09:45) RQ: without diving into locale stuff

(19:10:01) # Benutzer thorstenziehm__ ist dem Kanal #ooonlc beigetreten

(19:10:06) RQ: and then l10n teams could focus on more thorough tests on selected/available platforms

(19:10:34) Thalion72: rafaella / RQ: you are right in theory - but we already have prroven, that one single localized build can be broken even if all others are ok

(19:10:54) RQ: Thalion72, but do you need a localizer to test it for being broken?

(19:11:07) Thalion72: RQ: yes

(19:11:13) _nakatamaho_: yes

(19:11:23) RQ: do you need a person who knows lithuanian to tell that Ooo Lithuanian doesn't run or install?

(19:11:25) Thalion72: RQ: i have no problem, if he build is "announced" as untested

(19:11:34) RQ: me neither

(19:11:54) RQ: my main problem is that untested builds go to extended/localized/

(19:11:58) RQ: and aren't easy to find

(19:12:00) sophi: Thalion72: yes making the difference visible is important

(19:12:11) _nakatamaho_: right

(19:12:13) Thalion72: RQ: so - if you have testers .. do some minimal tests and release the builds

(19:12:15) RQ: oh, and they don't even go there. they are actually ONLY available from linux.cz

(19:12:34) _nakatamaho_: RQ: you can define how to test

(19:12:41) _nakatamaho_: to get approved

(19:12:45) Thalion72: if you have no testers: do no tests and post an notice, that untested builds are available and ask people for feedback

(19:12:49) RQ: Thalion72, well, I have one OSX/PPC tester, who has passed two testcases and hasn't tried to pass others yet

(19:13:07) RQ: but the rule is that I have to have at least two testers per platform to get approved

(19:13:08) RQ: isn't it?

(19:13:19) Thalion72: RQ: in this case you at least know, that the build can be installed and starts

(19:13:32) Thalion72: RQ: no ..it is up to you to decide

(19:13:40) RQ: Thalion72, oh that's better

(19:14:01) RQ: Thalion72, but I don't have to know Japanese to tell you if a Japanese build installs for me or not

(19:14:15) RQ: why do you need to know Lithuanian to tell the same about a Lithuanian build?

(19:14:49) Thalion72: RQ: can you identify if the Japanese build does display "get check chicks from here..."?

(19:14:58) Thalion72: (cheep chicks)

(19:15:04) RQ: no

(19:15:30) RQ: but a Japanese localizer on a different CPU can

(19:16:06) # thorstenziehm__ hat den Chat verlassen (Beenden: "Read error: 60 (Operation timed out)" ).

(19:16:10) Thalion72: RQ: once again .. we hat localized builds that failed on only one plattform and for only one localization

(19:16:25) RQ: Thalion72, they failed how?

(19:16:38) RQ: by telling to get cheep chicks or by not installing?

(19:16:58) Thalion72: by not installing .. but the point is, that they were "different"

(19:17:06) RQ: I see the poing

(19:17:32) RQ: but my point is that the STRINGS are the same, and even a machine could test whether BINARY code is OK or not

(19:17:53) Thalion72: RQ: how do you knowthat?

(19:17:57) RQ: you don't need a person who knows the language to fail an installer

(19:18:17) RQ: but you DO need a person who knows that language to test if the installer isn't telling you to get chicks

(19:18:44) RQ: but if it is, it's at least telling that on all CPUs in the same OS

(19:19:04) Thalion72: RQ: yes - I would not release a GErman build that has not been tested by someone who is able to read German

(19:19:16) Rail: but there are some locale related things which are not in strings

(19:19:17) RQ: so you can test on Windows/x86 to be sure that Windows/x64 isn't promoting chicks

(19:19:39) rafaella: Hoinestly, I think that it's the teams who should decide what works best for them in regard to testing...

(19:19:57) _nakatamaho_: RQ: you can define criterion

(19:19:58) rafaella: I think that we are here to share experience and best praactices

(19:19:59) Thalion72: RQ: for this example - yes .. because we use the same builds on Win 32 / Win 64

(19:20:07) RQ: well, then I'd go with differentiating OS's, not CPUs

(19:20:29) RQ: Thalion72, s/windows/mac/, s/x86/ppc/

(19:21:32) Thalion72: RQ: mac intel and mac ppc builds are taken from different build environments (at least the GErman builds) - so there *is* a good chance that they are different

(19:21:48) RQ: Thalion72, STRING different?

(19:21:51) Thalion72: yes

(19:21:57) _nakatamaho_: Can be

(19:22:04) RQ: hm

(19:22:21) RQ: how can that happen?

(19:22:35) _nakatamaho_: hope the source codes are the same :)

(19:22:40) Thalion72: RQ: anyway .. I agree with rafaella it is up to you to decide .. but you need to be aware of the risk

(19:22:44) RQ: aren't strings being frozen when releasing?

(19:23:07) Thalion72: RQ: yes .. but you just need to have a slightly broken checkout

(19:23:08) # thorstenziehm hat den Chat verlassen (Beenden: "Read error: 110 (Connection timed out)" ).

(19:23:17) RQ: hm...

(19:23:19) Thalion72: or oldfiles in the build tree ...

(19:23:46) RQ: hm...

(19:24:04) RQ: then I guess the build process should be improved to prevent that from happenning

(19:24:08) Thalion72: at the other hand - you never can check *all* before a release ...so there is still no guarrantee even if you test

(19:25:02) Thalion72: RQ: agreed .. but did you ever see a perfect software without errors ? (and the build process is not much more than a bunch of software)

(19:25:49) RQ: well, I'm a localizer at mozilla, and they have improved their build process last year :)

(19:26:11) RQ: to the point that all platform buildbots check out exactly the same sourcecodes

(19:26:15) rafaella: I would like to recap a bit and conclude the meeting, if that's ok with you

(19:26:21) Thalion72: RQ: take it th eother way .. why do you need to "release" someting, insteadof being honest and say "we have builds for you but these untested"

(19:26:49) RQ: Thalion72, I don't need to release, but I don't want people to have to look on FTP servers manually either

(19:27:05) # thorstenziehm_ hat den Chat verlassen (Beenden: "Read error: 110 (Connection timed out)" ).

(19:27:11) RQ: if we could put all builds for download, marking some of them as untested, that would be great

(19:27:16) Rail: you cannot check some OS dependant strings (Ctrl for PC, Cmd for Mac for example) if you don't have that OS

(19:27:27) Thalion72: RQ: so put the links on your website and mark them as untested / Release candidate (yet to be tested)

(19:27:34) RQ: but right now it's like all or nothing (either builds are released, or hidden)

(19:28:04) Thalion72: RQ: where do your user go for downloading OOo?

(19:28:06) RQ: Thalion72, I wish to see them on openoffice.org instead of some website that not many people know about.

(19:28:21) RQ: Thalion72, oo.o, i guess

(19:28:24) Rail: but I agree with RQ that we can publish untested builds on l10n sites as untested

(19:28:39) Thalion72: RQ: www.ooo or your native-lang page?

(19:28:42) rafaella: I need to go now, can some of you (Rail, Thalion72 or RQ, ...) send the link to the IRC log?

(19:29:15) Thalion72: hmm .. i have the log .. can put it to the wiki (at least i think so)

(19:29:21) RQ: Rail, btw I wasn't talking about testing on Linux, releasing on mac. I was talking about testing Mac/x86, and releasing Mac/PPC

(19:29:32) rafaella: If needed we can schedule another meeting on QA testing

(19:29:43) RQ: Thalion72, www.ooo, i guess

(19:30:00) sophi: rafaella: thanks and sorry for being late

(19:30:09) Thalion72: rafaella: at the moment it sems to be more about "how to make builds available for users" - even if untested

(19:30:11) RQ: rafaella, I don't have a log, sorry

(19:30:22) RQ: Thalion72, right

(19:30:26) _nakatamaho_: hope the source codes are the same :)

(19:30:34) _nakatamaho_: opps

(19:30:37) rafaella: Thank you for attending talk to you soon again!

(19:30:38) RQ: I think we could at least have all builds in /localized

(19:30:41) RQ: even untested ones

(19:30:46) Thalion72: RQ: hmm ... we promote de.openoffice.org .. and we are free to put there what we want

(19:30:46) # rafaella hat den Chat verlassen (Beenden: ""Leaving"" ).

(19:30:51) _nakatamaho_: rafalella: let's have an another meeting :)

(19:30:52) RQ: (they could have UNTESTED in filenames, or whatever)

(19:31:08) Thalion72: RQ: they would have RC ;)

(19:31:16) RQ: that's fine

(19:31:21) Rail: +1

(19:31:33) RQ: but as a mirror admin I know that it's pissing to get my builds from all around the www

(19:31:39) _nakatamaho_: all: I'd like to go. 3:30 am at Japan...see you

(19:31:47) Thalion72: _nakatamaho_: CU

(19:32:03) RQ: _nakatamaho_, good night

(19:32:04) sophi: _nakatamaho_: see you and thanks :)

(19:32:05) RQ:  ;)

(19:32:18) Thalion72: RQ: problem is that we migh blow up the mirrors, if we put all RC to /localized

(19:32:37) Thalion72: we once had them there .. but run out of space

(19:32:50) RQ: Thalion72, back when my mirror had only 54GB of space, I was concerned about it

(19:32:50) Thalion72: therefore they are now on /extended

(19:32:58) khirano: Good night, Maho

(19:32:59) RQ: now you can buy HDD's with terabytes

(19:33:23) Thalion72: RQ: "you" yes .. but this should be discussed with the mirror admins - notwith me ;)

(19:33:27) RQ: Thalion72, and mirror admins don't have to mirror ALL localized if they can't

(19:33:49) RQ: back in that time I was mirrorring only selected dirs

(19:33:57) Thalion72: I can oly tell that it is not easy to get answer from mirror admins (at least this is what I have been told by Florian)

(19:33:58) RQ: but I myself decided what to select

(19:34:08) RQ: heh

(19:34:09) RQ:  :)

(19:34:42) RQ: it's because the matters that are currently being discussed on mirrors mailing list aren't really important

(19:34:47) RQ: (for us)

(19:34:55) Thalion72: RQ: could you check this with the distribution project .. if it was possible to get locliced RC back to /localized ?

(19:35:05) RQ: you mean with the mirrors ?

(19:35:09) Thalion72: yes

(19:35:12) RQ: ok

(19:35:22) khirano: Good :)

(19:35:36) Thalion72: I think, we have two points here:

(19:35:56) Thalion72: - how to distribute untested builds

(19:36:02) # _nakatamaho_ hat den Chat verlassen (Beenden: ""Xirc v0.9.9 - MacOSX"" ).

(19:36:16) Thalion72: - how to get people interested in testing such builds .. so that they can be "released"

(19:36:41) khirano: Thalion72: Right.

(19:36:49) sophi: Thalion72: the points are joined, it's true that no testers will come if they don't find builds

(19:36:58) RQ: exactly!

(19:37:08) RQ: Thalion72, good summary! :)

(19:37:13) Rail: +1

(19:38:01) Thalion72: RQ: but I hope, the option to easily distribute untested builds will not stop you from testing and releasing ;)

(19:38:11) RQ:  :)

(19:38:15) RQ: sure

(19:38:48) khirano: So we will discuss these joined two points in the next meeting, right?

(19:39:00) RQ: actually, if we made them more visible, we could even employ some feedback right on the web

(19:39:20) Thalion72: khirano: 1st point is something for the distribution project .. I'd aks RQ to handle this

(19:39:28) RQ: like, a person could come and mark their build as 'working' or 'not working', for example

(19:39:48) khirano: Thalion72: sure :)

(19:39:59) Thalion72: RQ: hmm ... this was one of the ideas behind QATrack

(19:40:21) Thalion72: maybe the barriers are still to high

(19:40:23) RQ: this would also mean that people with lower expectations would mark builds as working more often than those with higher expectations :)

(19:40:33) RQ: and I think it's good

(19:40:48) Thalion72: actually e (german team) collect such feedback via mailing list

(19:41:25) sophi: RQ: I think mailing lists is better because you need feedback if the build doesn't work, you need to know why

(19:42:00) RQ: but that's like extended feedback :)

(19:42:13) RQ: and what I was talking was something very basic

(19:42:55) Thalion72: if the basic feedback is "it doesn't work" you urgently need more extended feedback :)

(19:43:03) RQ: sure :)

(19:43:14) RQ: but if it is "it works", then at least you can say that to others :)

(19:43:22) RQ: and it's better than nothing :)

(19:43:23) Thalion72: in general the basic feedback is "hmm .. well .. it works .. but ..."

(19:43:36) RQ: well, it's always like that :)

(19:44:35) Thalion72: ok .. anything open to discuss ?

(19:45:28) Thalion72: seems not - soI'll thake the logand put it to the wiki (and try to draft some minutes)