IRC log of #schooltool for Wednesday, 2007-01-03

*** jfroche has quit IRC00:04
*** mortons has quit IRC00:25
*** didymo has quit IRC00:27
*** aclark has quit IRC00:27
*** didymo has joined #schooltool00:28
*** aclark has joined #schooltool00:28
*** aclark has quit IRC00:33
*** didymo has quit IRC00:33
*** didymo has joined #schooltool00:34
*** aclark has joined #schooltool00:34
*** aclark has quit IRC00:43
*** didymo has quit IRC00:43
*** didymo has joined #schooltool00:43
*** aclark has joined #schooltool00:43
*** mattva01 has joined #schooltool00:49
*** didymo has quit IRC00:51
*** aclark has quit IRC00:56
*** aclark has joined #schooltool00:56
*** DeathOmen has quit IRC01:29
*** mattva01 has quit IRC01:43
*** lisppaste5 has quit IRC02:23
*** lisppaste5 has joined #schooltool02:27
*** ddaa has quit IRC05:25
*** th1a has joined #schooltool05:30
*** didymo has joined #schooltool05:34
*** Bhaskar has joined #schooltool05:47
Bhaskarthla: goodmorning sir05:50
Bhaskarthla: any new news about schooltool05:52
th1aBhaskar: I am, in fact, writing an email on your behalf at this very moment.05:53
Bhaskarthla:well sir05:54
*** didymo has quit IRC06:50
Bhaskarthla: are u BG sir07:22
th1aBhaskar: Am I BG?07:38
Bhaskarthla:well07:40
Bhaskarthla:i am translating schoolbell now07:41
th1aBhaskar: You really don't want to bother with SchoolBell.07:41
Bhaskarthla: sorry ? you mean...07:42
th1aSchoolTool.07:43
th1aFocus on SchoolTool.07:43
Bhaskarthla: well our first priority is for schooltool07:44
Bhaskarthla: i have translate schooltool  and download .op and .mo file from launchpad, then i  make update-translation but i could not get success to view school tool in nepaly completely, only partially , why07:45
th1aAre you running SchoolTool from a svn checkout?07:53
th1aOr from apt-get?07:53
Bhaskarthla:ya07:57
th1ayes, apt-get?07:57
Bhaskarthla: sudo apt-get install subversion08:00
Bhaskarthla: sudo apt-get install build-essential08:01
th1aBhaskar: But you're running SchoolTool from subversion?08:02
Bhaskarthla: i have done all this command08:02
Bhaskarthla:no no08:03
Bhaskarthla:i run sudo apt-get install build-essential08:13
Bhaskarthla:no use of subversion08:14
th1aIgnas said he thought you'd done the translation for the current development version of SchoolTool (from subversion) and tried to apply it to the version of SchoolTool in Ubuntu, which is much older.08:15
Bhaskarthla: i use schooltool-2006.0-alpha208:20
th1aHm...08:22
th1aYou should probably use a subversion checkout.08:22
Bhaskarthla:please tell me for this package which template be translate in launchpad08:23
th1aAs far as I know, you should use the 2006 template on LaunchPad and a subversion checkout.08:23
th1aThat should work, and if it doesn't, we should fix it.08:24
Bhaskarthla:i use template as08:30
BhaskarSchoolTool Series: 200608:30
BhaskarTranslation template "schooltool"08:30
th1aThat's the current one, as far as I know.08:30
Bhaskarthla: i think it should work properly for schooltool-2006.0-alpha208:31
Bhaskarthla: suggest me how can i do08:32
th1aMy suggestion is to try a subversion checkout.08:33
th1aJust in case.08:33
Bhaskarthla: i think i can use template:08:38
BhaskarSchoolTool Series: main08:38
BhaskarTranslation template "schooltool"08:38
Bhaskarthla: OR08:40
Bhaskarschooltool in Ubuntu Dapper08:40
BhaskarTranslation template "schooltool"08:40
Bhaskarithla: i use Ubuntu 6.06 LTS08:41
th1aBhaskar: OK.  I have no idea.08:42
th1aI am monolingual.08:42
th1aI don't understand how translation works at all.08:42
Bhaskarthla: please share these events with your team08:43
th1aOK... I'll see if I can get someone to help you out.08:44
Bhaskarthla: Please08:44
Bhaskarthla: there is confusion in launchpad.net , for which template should be used08:45
th1aYes, I see that.08:46
Bhaskarthla: i think it should be simplified......08:46
Bhaskarthla: i think backward compactibility should be maintained , i mean newer version of template should support for older one package08:47
th1aAt this point, we really don't know what the problem is.08:49
Bhaskarthla: please discuss with your team, there are many confussion in launchpad.net for translation template08:50
th1aBhaskar: Is it 12:36 PM where you are?08:51
Bhaskarthla: me in office see www.mpp.org.np08:52
Bhaskarthla: in Nepal it is mid day08:52
Bhaskarthla: you can see www.nepalinux.org08:53
Bhaskarthla: we developed Nepalinux08:53
th1aAh.  I see.08:53
th1aThanks.08:53
Bhaskarthla: we are the team, that starting the revolution in opensource as well localization in Nepal08:55
Bhaskarthla: we developed Nepali unicode, Nepali SMS, Nepali text to speech, Nepali dictionary and many more work08:56
Bhaskarthla: our latest progress is on : schooltool08:57
Bhaskarthla: I am here for Schooltool localization, customization and research for unicade08:57
Bhaskarthla: i am  young IT Engineer08:58
th1aI see you've got a lot going on.08:59
Bhaskarthla:well your suggestion and enouragement is valuable to us for further going09:08
Bhaskarthla: have you made schooltool mannual, we are looking to localize for this mannual09:12
Bhaskarthla: please send me schooltool mannual!09:13
*** Aiste has joined #schooltool09:15
BhaskarAiste: have u know schooltool09:23
AisteBhaskar: please stop harassing members of this chat09:26
AisteIgnas clearly explained what you need to do if you want to ask further questions09:27
BhaskarAiste: how can i meet with him09:28
Aistewhy do you need to meet with him?09:28
BhaskarAiste: my intension to encourage the people ,not harass, i think we can share our problem on schooltool in this forum09:29
Aisteyes, but in this case you are harassing09:30
AisteIgnas has a lot of work to do on Schooltool and he can only participate in constructive discussions09:31
Aistenot the ones like - we need to increase the community, yeah yeah type09:31
Aisteyou can share the problems, but it seems that the cultures are quite different and what you consider amiable conversation other people perceive as too invasive and annoying09:32
BhaskarAiste: i am also have a lot of work on schooltool, i also trying in constructive work09:33
Aisteso, I'd rather you send emails to the mailing list with constructive questions and suggestions and not try to talk to all the possible people on irc09:33
BhaskarAiste: thanks for your suggestion09:34
Aisteyou are welcome09:36
BhaskarAiste: i am engineer, donot waste time in chating but want to utilise in some context,09:49
*** mgedmin has joined #schooltool10:37
*** Bhaskar has quit IRC10:41
*** Bhaskar has joined #schooltool10:43
*** jfroche has joined #schooltool11:25
*** jinty has joined #schooltool11:34
*** thisfred has joined #schooltool11:48
*** Bhaskar has quit IRC12:42
*** Ninno has joined #schooltool13:12
NinnoHi13:12
*** alga has joined #SchoolTool13:42
*** ignas has joined #schooltool13:50
*** jfroche has quit IRC14:06
*** ddaa has joined #schooltool14:21
*** ddaa has left #schooltool14:21
*** ddaa has joined #schooltool14:21
*** joerg has joined #schooltool14:32
*** joerg_ has quit IRC14:45
*** Aiste has quit IRC14:49
*** Ninno has quit IRC15:04
*** Aiste has joined #schooltool15:30
*** karstix has joined #schooltool15:46
*** karstix has left #schooltool15:48
*** Ninno has joined #schooltool16:19
*** Ninno has quit IRC16:47
*** thisfred has quit IRC17:58
*** wdickers has joined #schooltool17:58
*** mattva01 has joined #schooltool18:00
*** aelkner has joined #schooltool18:02
aelknerignas: ayt?18:04
ignasyes18:04
aelknerhave you been looking at chesty, the tinyzis agent lib?18:06
ddaaignas: how late are you staying online today?18:08
ignasaelkner: no, not really18:08
ignasddaa: hmm, at least a couple of hours18:08
ignashow much do you need ?18:08
aelknerok, thatnks anyway.18:08
ddaaI would like to talk with you later about your use cases.18:09
ignasok18:09
ddaaWe are writing a spec right now about the different ways to subscribe to a branch.18:09
ddaaSo if we can capture precisely what you need, we'll probably be able to implement exactly that.18:10
ddaaHow cool is that?18:10
ddaa(implement in Launchpad)18:10
ignasneat :)18:12
ignaspremium service ;)18:12
ignasddaa: maybe the proposal is available somewhere so i could read what's already in there?18:13
ddaaunfortunately, it's on a Canonical-only wiki18:13
ddaabut anyway, I think it's better not to taint you18:13
ddaajust talk about use cases18:14
ddaabut right now I need to go out buy some camping gear18:14
ignasok18:14
th1aaelkner: What's up?18:23
aelknerHey Tom.18:24
aelknerWill and I are looking at the latest on chesty trying to figure out what you're up to.18:24
aelknerAfter doing the svn up, I tried to run python test.py, but it complains about tlslite.api.  Could you please provide instructions on how to get test.py to run?18:25
th1aOh yeah.18:26
*** mattva01 has quit IRC18:26
th1aYou need to install tinyzis, which now adds tlslite to your site-packages.18:27
wdickerscould you give us the location of the latest tinyzis repository?18:28
th1aUh... has it moved?18:29
*** karstix has joined #schooltool18:29
wdickersno, but the url would be useful, especially chesty's18:30
th1aDoesn't svn co https://svn1.hosted-projects.com/sif/tinyzis/trunk/tinyzis work?18:30
aelknerI did svn co https://svn1.hosted-projects.com/sif/tinyzis which worked18:30
wdickersthanks, and chesty?18:31
aelknerShould I do the trunk only, Tom?18:31
th1aaelkner: It pretty much depends on how much extra stuff you want lying around.18:31
aelknerOk.18:32
th1asvn co https://svn1.hosted-projects.com/sif/pyagentlib/trunk/chesty18:32
wdickersgot them, thanks18:33
wdickersI have a question about the sendAck() in agent.py. All the other send methods invoke postandprocess() but sendAck() does not, why?18:34
aelknerTom, could you update the tinyzis README.txt with more instruction for installing everything?18:34
aelknerRight now it says, "once you have everythong installed...", but that's assuming the developer knows how to do that.18:34
th1aBecause sendAck() doesn't expect a response, I think.18:36
wdickersah, that makes sense18:36
aelknerpython setup.py --help-commands yields a long list of commands, but which one needs to be done.18:36
th1asudo python setup.py install18:36
aelknerOk, so that's what needs to go in the README.18:37
wdickersI'm afraid I'm out of time. I will ask Mr. Elkner to install python-xml and tinyzis on maddog, is that all that needs to be done to get it working?18:37
th1aThat should do it.18:38
aelknerYou should refer to the README after doing a svn up the next time you log in.18:38
wdickersalright, see you two tomorrow18:38
aelknerSee you.18:38
*** wdickers has quit IRC18:38
aelknerth1a: are you updating the README.txt file in the tinyzis directory as well as the chesty directory so that we know how to poceed?18:40
th1aI will do it.18:40
aelknerThanks.18:41
aelknerSo after doing the tinyzis install, I can get chesty's test.py to run, but it only outputs two messages.  Is that right?18:43
th1aYes.18:43
aelknerOk.  python ftest.py yields an error message: error: (111, 'Connection refused')18:45
aelknerFailed example: agent.sendRegister()18:45
th1aYou have to start up a tinyzis instance manually.18:46
aelknerSo you're going create a README.txt for chesty as well, right?  Please include instructions like that.18:47
th1aYes, aelkner.18:48
aelknerOk.  I'm happy now.  Having run tz-run.py, going back to chesty, ftest.py works.18:51
*** Ninno has joined #schooltool18:57
Ninnohi18:57
*** aelkner has quit IRC19:00
*** Ninno has quit IRC19:09
*** jinty has quit IRC19:47
*** karstix has left #schooltool20:19
*** Aiste has quit IRC20:41
ddaaignas: ping20:46
ignasddaa: pong20:47
ddaaSo... I'd like to discuss about what you're going to need from bzr20:47
ddaato support your workfow20:47
ignasthe mailing/coordination part?20:48
ddaaAt some point we're going to talk about emails, but I'd like the discussion to be focused on _what_ you do, rather than _how_ you think I could be done with bzr.20:48
ignasok20:48
ignasas all the branches are in the same svn repository20:49
* ddaa listens20:49
ignasevery checking triggers an email to the schooltool-checkins mailing list20:49
ignasreviews are done after the commit by replying to these emails20:49
ddaaso every commit on every branch goes to schoooltool-checkin@, with the diff and commit message?20:50
ignasyes20:50
ignasall developers are subscribed to the list20:50
ignasso everyone can see what other programmers are working on20:50
ignascommits to trunk also trigger buildbot to run all the tests on an up to date trunk20:51
ddaahow does the triggering happen?20:51
ddaais that a svn post-commit hook, or it's triggered by commit emails, or somehing else?20:51
ignaspostcommit hook uses XMLRPC to notify buildbot that there are new changes20:52
ignasbuildbot retrieves a more or less clean checkout and runs all the tests on it20:52
* ddaa waits for more or an EOF marker20:53
ignasi can EOF now, or tell you about active branches20:54
ignas:)20:54
ddaatell me about active branches, please20:54
ddaawhat is an "active branch"?20:54
ignasbranches that had changes in recent 2-3 months20:54
ignasand will be either worked on or merged to trunk20:55
ignassooner or later20:55
ignasat the moment we have 3-4 active branches, new-style-navigation  (hasn't been touched for a while is very different from schooltool trunk it was branched like 4 months ago and has a lot of new code in it)20:55
ignaslyceum branch - has different config files and a separate src/lyceum directory with lyceum specific code20:56
ignasjfroche's branch (can't recall the school name) - is more or less trunk, but has a small feature being worked on at the moment20:56
ignasand trunk - i am developing changes useful to all schooltool users in it20:57
ddaaThe vocation of an active branch is to be merged to trunk?20:57
ignasnew-navigation - will be merged to trunk (in 6-12 months)20:58
ignaslyceum and the other school might have some features developed for them that we might want to have in trunk20:59
ignasand everything on trunk should get to lyceum and the other school20:59
ignasbut full school branches will not get merged to trunk, only some features might20:59
ddaaSo, the "full school" branches are effectively friendly forks?21:00
ignasyes21:00
*** iatrogenic has joined #schooltool21:00
ddaaignas: I do not want to interrupt your flow, so tell me when you expect me to prompt you.21:01
ignasyou can do that now21:01
ignas:)21:01
ddaaSo, all the people who are actively working on schooltool have svn commit access? Or is there development activity going on outside of the svn?21:02
ddaa(like developers who submit patches to a mailing list)21:02
ignasno, we don't have outside commiters, our QA is too harash for people to work on schooltool without our help21:03
ignaspatches have to be worked on by at least one developer 90% of the time21:03
ignasfixing tests21:03
ignasfixing the bug properly21:03
ignasfixing all the instances21:03
ignasetc.21:03
ddaaOkay, I see how DVCS can lower your barrier to entry.21:03
* ddaa reads backlog to find more questions21:04
ddaaSo, at the moment, once there is any activity at all, everybody knows through the checkins ML.21:05
ignasyes21:05
ignasat least i am reviewing every single checkin21:05
ignaswith only 1 other developer it's fairly easy21:05
ddaaI guess that one thing you expect from switching to bzr is to have more people doing checkins.21:05
ignasi am not sure we will get anyone :) , Zope3 + 90% unit test coverage + code quality standards are making it very difficult to work on schooltool :/21:06
ignasfor other people that is21:06
*** iatrogenic has left #schooltool21:06
th1aignas: We'll work on evangelism after we've got some schools running.21:06
ddaaSo, say you get two other people checking in random broken stuff (at first, until they learn) on their own branch. You would like these commits to go to the same mailing list?21:07
ignasyes, possibly, i want them to have that option21:07
ignaslike - registering their branch21:07
ignasso i would get emails21:07
ignasif they want feedback21:07
ignasalthough they might want to work alone ;)21:08
ddaaYou or everybody reading the mailing list?21:08
ignashmm, a good question, i have learned a lot of things by reading mgedmin's comments on checkins of other programmers too21:08
ignasso having it on a mailing list might be a good idea21:09
ignas.21:10
ddaaSay somebody makes a change _for_ trunk.21:10
ddaaYou review it, and say "good".21:10
ddaaThen it gets through buildbot.21:10
ddaaIf buildbot finds some errors21:10
ddaaWhat happens next?21:10
ddaaBTW, did I describe the process for committing to trunk properly?21:11
ignasnope21:11
ignassome one commits to trunk21:11
ignasbuildbot starts chugging21:11
ignasbuildbot fails, sends an email that peter just broke the trunk21:12
ignasi wake up and review his checkin, comment on it, and yell at him for not running tests before commiting21:12
ignasthen I either fix it, or give him the possibility to fix it himself21:13
ddaaWould you prefer if it was not possible to commits something to trunk that breaks the test suite?21:13
ddaaOr do you want to allow broken trunk commits?21:14
ignaswith bzr in place - it would be better to avoid breaking trunk21:14
* ddaa goes back to the scrollback21:15
ignasbreaking trunk only happens when you forget to svn add something so bumping such checkins is a good idea21:16
ddaaDo the "active branches" have a special significance to the toolchain, or are they just a conventional way to refer to a set of distinguished branches?21:17
ddaaor something else?21:17
* ddaa slaps himself for asking binary questions21:17
ignasno they are not a part of the toolchain at the moment21:18
ignasthough i would like to have buildbot running on lyceum-branch21:18
ignasas a safeguard to avoid deploying broken things21:18
ddaaanything preventing it?21:18
ignashmm, with svn - i don't really know, jinty is responsible for everything at the moment21:19
ddaaBTW, what is the role of jinty now?21:20
*** jfroche has joined #schooltool21:20
ignasjinty is the administrator of the server all our infrastructure lives on21:20
ignasand would be the release manager if we were making releases21:20
ddaadoes he have a programming role?21:21
ignashe does some tweaks to schooltool Makefiles, configuration files, some source code if it is required to make releases work21:22
ddaaschooltool trunk, right?21:22
ignasreleases were being done from schooltool trunk21:22
ignasso yes21:22
ignassome of his tools live in the svn repository too21:23
ddaathey are part of buildbot, or they happen to share the repository out of convenience?21:23
ddaaor something else...21:23
ignasconvenience21:23
* ddaa slaps himself again21:23
* mgedmin remembers the spanish inquisition sketch21:24
ddaawhat kind of tools?21:24
ignasdebian-packaging, release branching + tagging + packaging scripts21:24
ddaaok21:25
ddaaI think I have a better idea now of how your are working.21:25
* ignas too21:25
ignas:)21:25
* ddaa inserts a bookmarks for logs sake ========================21:26
ddaaI think the first thing that needs deciding is where the schooltool branches will be hosted.21:27
ignasoptions? :)21:27
ddaaRequirements, first :)21:28
ddaaSo, branches like lyceum or otherschool, or jinty's stuff are just committed to by a single person. Right?21:28
ignasyes21:29
ignashmm, i want branches that i am getting checkin emails about visible and accessible on the internet21:29
ddaaSure.21:29
ddaaBut the trunk branch (and maybe other branches in the future) are currently committed to by multiple people.21:29
ddaaRight?21:30
ignasyes21:30
ignasi can always tweak jintys release scripts, if it is needed21:30
ignasi can finish up jfroches work on his branch so i could merge it to trunk21:30
ignasetc.21:31
ddaaCan "tweak jintys scripts" or "finish up jfroches branch" be done as conveniently using separate bzr branches?21:32
ignasyes i think so21:32
ignascheckout, fix, publish, ask them to merge/pull if they want21:32
ddaaWhat I was thinking.21:32
ddaaSo... generally there are three options for hosting branches.21:33
ddaa1. Wherever on the internet. For branches committable by a single person. Launchpad already has a robust branch mirroring facility, so as long as it's online like half the time for Launchpad to get, it's enough.21:34
ddaa2. Uploaded on Launchpad. That works for branches committable by a single person, but also allows multiple people to commit to a branch.21:35
ddaaThe main drawback is that creating a new branch on Launchpad currently requires a complete upload of all the branch data.21:37
ddaa(we plan to fix that using smart server smartness, but it's not yet being worked on)21:37
ddaa3. Uploaded on a non-launchpad central server. A simple way to do it is to create accounts people can sftp into. You can set the shell to scponly to prevent these accounts from ever getting an interactive shell.21:38
ddaaIn this case, you can use things like unix groups to allow select people to write to repositories not owned by their user.21:39
ddaaNote that the presence of a repository in a remote server is only useful to speed up uploading. It has no effect on download speed (you want a local repository for that).21:40
ignashmm, 1 + 3, trusted developers that are a part of the release cycle or just don't have their own web space have accounts21:40
ignastrunk + historical branches kept on schooltool server21:41
ignastrunk + developer branches + third party developer branches mirrored in launchpad21:41
ddaaYou can also mirror historical branches to launchpad. It acts as a free backup service, too :)21:42
ddaaNote that you currently cannot easily delete branches from Launchpad.21:43
ddaaBut branches on Launchpad have a "status", if it's Merged or Abandoned, the branch does not show up in listings by default.21:43
ignasoh21:43
ddaanew feature, just rolled out :)21:43
ddaacourtesy of thumper21:44
ddaahttps://code.launchpad.net/bzr/21:44
* ddaa makes another bookmark ==============================21:45
ddaaDriving buildbot21:46
ddaaI do not think you can conveniently use the post-commit hook approach with bzr ATM21:46
ddaasince post-commit hooks are run on client21:46
ignaswell, you could do it on push, but hmm, push is not something you can easily prevent ...21:47
ddaapush does not run any post-commit hook21:47
ddaabut the recommended way to work with multi-committers branch is to use checkouts21:48
ddaawhere you do not normally run push21:48
ignasbbot only works if some robot is doing the merging to the guarded branch ...21:48
ddaaNot necessarily...21:48
ddaaI see three approaches here:21:48
ddaa1. as you said, some robot runs server side to do the merging21:49
ddaapqm would be a very good fit for your requirements because it effectively prevents people from comitting revisions that fail the test suite21:49
ddaabut it's kind of a pain in the ass to administer, so let's keep it in the later basket, and instead find ways to drive buildbot21:50
ddaa2. a custom bzr plugin used by trunk comitters. It would quick buildbot in the xmlrpc when doing a remote commit on a checkout of trunk.21:51
*** alga has quit IRC21:51
ddaathe annoying thing is that it relies on trunk committers a. only doing commit to trunk, not push b. have the plugin installed and working21:52
ddaamight be easy to fail this just by mistake21:52
ddaa3. have cronscript that reads the checkins mailing list and quick buildbot when it sees an email about a code change on trunk21:53
*** jfroche has quit IRC21:54
ignashmm21:54
ddaathat would make the checking ML a single point of failure, but in case of disaster you can still run buildbot manually21:55
ddaaand I think it would be simple enough to hack using stock command line tools21:55
ignasML is on schooltool.org, so cron script would just look at real files21:56
ignasso i have a published writable repository bzr.schooltool.org/lyceum/21:56
ignasi bzr get st.org/lyceum21:56
ignasbzr merge my-new-feature to it + commit21:56
ignasiirc changes are not in schooltool.org yet21:56
ddaaactually "bzr checkout st.org/lyceum"21:57
ignasoh21:57
ddaathen your changes are uploaded on commit unless you specify "--local"21:57
ddaaignas: go on21:58
ignasso bzr merge + commit21:58
ignaseverything gets published21:59
ignasemails get somehow sent21:59
ignascronscript gets the email and starts buildbot21:59
ignasi get a warning21:59
ignassounds good21:59
ignasone of the scary things about bzr is that I don't know how to fix a branch that got misused ...22:00
ddaadepends of what kind of misuse you are thinking.22:00
ddaaand what you mean by fixing22:01
ddaaFirst, can we call the buildbot thing a solved problem for now?22:01
ignaswith svn it's all well clear - svn keeps all the history, even messy one22:01
ignasbuildbot - yes i think so22:01
* ddaa makes a bookmark ==============================22:01
ignasgot an unclean changeset merged into22:02
ddaaignas: with bzr you can do it the svn way22:02
ignasthen god a clean changeset for another feature22:02
ddaabzr merge -r-2..-322:02
ddaathat will apply the reverse of the change introduce in the penultimate revision22:03
ignasand even if i will fix the feature in the feature branch - it will be well, less nice in trunk than if one would get it directly22:03
ignasoh, so i revert a patchset then remerge the full/complete one22:04
ddaanot quite, reversion has somewhat surprising semantics in bzr. It's related to the lack of cherrpick tracking22:05
ddaaif you commit that reversion that will say "I merged the changes in that revision, and applied additional, unrelated, changes that reverse them"22:05
ignas?22:06
ddaaif you merge again the branch that was merged in the revision you reversed22:06
ddaabzr will only try to get the _new_ changes22:06
ignasoh22:07
ignashmm22:07
ddaaBut there's another way22:07
ddaaactually several other ways...22:07
ddaalet's start with the one that's simplest to explain22:08
ddaayou can "bzr uncommit -r-3", that will uncommit to revision -3 (antepenultimate)22:08
ddaathen "bzr revert" to discard those changes from the working tree22:09
ddaaThen you can "bzr merge" the branch with the good fix that you want to keep.22:10
ddaaSince you uncomitted the merge of the broken feature, next time you try to merge it, after the fix, it will get all the changes.22:10
ddaaignas: makes sense?22:10
ignasyes22:11
ignasa nice bandaid if I merge+commit something to trunk by accident22:11
ddaaBut uncomitting multi-committers branch is frowned upon, and does not work too well.22:11
ddaaTake my word for it, you only want to uncommit on a single committer branch. There is a good reason for this, but it would distract us from the discussion at hand.22:12
ddaaIn a multi-committer case I think the right way to pull back a change is:22:13
* ddaa thinks about it22:13
ddaabzr folks are not big on multi-committers branches usually, got to stop and think :)22:14
ignasi don't think schooltool will be big on that22:15
ignasi mean with bzr - trunk will have to be the sanctuary of stability22:15
ignaswith preferably only nice full merges22:15
ddaabasically, you need to reverse the change with a reverse-cherrypick as previously, and this revision has to be merged into the feature branch without the tree change (effectively reversing the reversion), then the fixes are committed on the feature branch.22:16
ddaawhich blows up the annotation for those changes22:17
ddaaand is just messy22:17
ignasouch22:17
ignasnot a thing i'd want on trunk that will probably be merged from22:18
ddaaWe'll talk about why uncommit is bad on a multi-comitter branch another.22:18
ddaaignas: it's perfectly okay22:18
ignasor maybe trunk should be something one merges from, used only to make releases22:18
ddaait sounds complicated, and it's a bit of gymnastic, but it's not causing any harm, except from losing the annotation for the lines changed by the reversion.22:19
ddaathey will appear to come from the revision that reverses the reversion.22:19
ddaaSo, we'll talk about the uncommit-on-multi-committers another day.22:20
ddaaWant to finish this discussion first.22:20
ddaa=================================22:20
ddaaSo... emails :)22:21
ignasok22:21
ddaaBranches mirrored or uploaded on launchpad will be able to send email notifications.22:22
ddaaFor the code changes.22:22
ddaaBut also for the "details" changes.22:22
ddaaThings like the branch name, description, owner, whiteboard, status...22:22
ddaawhich are launchpad metadata22:22
ignasnot a problem really22:23
ddaaDo you want to receive emails about changes to this metadata to the checkin mailing list?22:23
ignasi don't think i care about it that much :)22:23
ignasthe metadata22:23
ddaaDo you want NOT to receive emails about those changes on the ML?22:24
ignasno difference to me at the moment22:24
ddaaOkay, at the moment we are going with "if you get diffs, you get metadata updates as well".22:24
ddaamakes the code and the UI simpler22:25
ignas:)22:25
ignasnow the question is - to whom ?22:27
ddaaSo for a subscribed (more later) branch, one email will be sent for each new revision.22:27
ignasschooltool-checkins has no account on launchpad ...22:28
ddaaYep.22:28
ddaaThere's a schooltool developers team, I guess?22:29
ignasyes22:29
* ddaa checks the spec22:29
ddaaWhat is currently proposed is associating an email to the product22:31
ddaaand maybe a subscription type as well22:31
ddaaso, this email would have to validated and owned by the product owner (the schooltool team I guess)22:31
ddaaIn the case of schooltool, mails about changes on all branches registered on the schooltool product could be sent to this email.22:32
ignasvalidation on the public mailing list will look funny :)22:32
ddaawell... it's a bit funny... but if the list is not public-postable, the validation message would be caught by the moderator.22:33
*** aclark has quit IRC22:34
ddaaAnd public-postable mailing lists are an endangered species because of spam nowadays22:34
ignastrue22:34
ddaaignas: so, do I understand well that you would like to receive emails about all and every branch registered on the schooltool product?22:35
ignasyes22:35
ddaaThere's a new edge case with commits email now.22:35
ignasat least at the moment, it might change when we'll get these 200+ developers attracted to the product :D22:35
ddaaWhen a branch is first registered, you do not want 6000 emails for all the revisions in the initial history :)22:36
ignasddaa: no, i kind of want 1 email that tells me that the branch was registered22:36
ddaaso I think in this case, a single "A branch was registered" email would be sent, without any diff.22:36
ddaaThe current proposal says: one email per new commit22:37
ddaathat means that if someone does a bunch of commits on his branch, and pushes them, you'll get a bunch of emails.22:37
ignasat the moment it will be good enough22:38
ignasi can gmail tag branches even :)22:38
ddaaIn the case where the push actually removes revision from the history, there would be a single "5 revisions removed" email, presumably with a diff22:38
ddaaignas: what would you find better?22:39
ignasi don't know, i have yet to see the trafic generated by this22:39
ddaaignas: for tag branches, you probably want to do merges anyway22:39
ddaaThere would probably be a limit on the size of the diff included in an email.22:40
ignasat the moment merges are represented by 1 huge email instead of 5 small ones22:40
ignasddaa: hmm, that might be a problem really22:40
ddaaignas: that would be the same22:40
ddaaa commit is one new revision in the branch history22:40
ddaait's many new revisions in the ancestry22:40
ignasi had some 100 kb diffs commited22:40
ignasand these had to get at least some review ...22:41
ddaaand you want to read that sort of thing in an email client?22:41
ignasmutt handles it :)22:41
ignasand ok 100 kb were not reviewable22:41
ignasbut 20 kb is common22:41
ignas:/22:41
ddaaI was thinking limit in number of lines22:42
ddaaso 100k should be about 20k lines?22:42
ignasthese are when you merge an old branch22:42
ignashmm22:42
ignaslet me check real sizes22:42
ignasor translations from rosetta22:43
ddaahu... presumably translations from rosetta are already reviewed, in Rosetta, by the time they hit svn...22:43
ignasyes22:43
ignasi know22:43
ddaaokay... so what you be the upper limit for a diff you are willing to receive by email?22:44
ddaa(and that other checkins subscriber are also willing to get)22:44
ignas200 kb i'd say22:45
ignas+ a link to a checkin diff on the web22:45
ddaaok22:45
ddaaneed fancy colors in the checkin diff?22:46
ddaaor is it just for reference? (e.g fancy colors provided by favourite text editor)22:46
ignasno not necessary, i want it accessible to be copy pasted22:46
mgedminfancy colors on the web == good22:47
ddaaSo... I think we'll need a setting to allow you to increase the max size of diffs sent by email.22:47
mgedminfancy colors in the email == no need22:47
ddaaRegarding the diff on the web, we have the basic infrastructure for storing millions of large files like this, but I do not expect it will happen right away.22:48
ddaaWe do want to have a real branch browser for launchpad (that would be a better way to provide the diffs), but that seems to be a significant challenge of its own.22:49
ddaaFor buildbot22:50
ddaawe'll probably need to add some interesting X-something headers to the email22:50
ignasor we'll do some parsing :)22:51
ddaalike X-Launchpad-Branch: ~schooltool/schooltool/trunk22:51
ignasthe way pov activity (timelog) mailing list works22:51
ddaawhat do you mean?22:51
ddaamaybe as well X-Launchpad-Branch-Change: [code] [details]22:52
ddaato avoid kicking buildbot if something changes the description of the branch22:52
ignaswe have like 5 different weekly report formats (different employees different habbits) and an ad-hoc system that parses out the date, hours worked etc ...22:53
ddaaokay22:53
ignasbut yes - X- headers would bi nice22:53
ignasfor everyone else too :)22:53
mgedminI imagine people might want tags in the subject22:53
ddaayup, one thing people like with malone is that it's got very useful headers to help sort the globs of bugspam it generates22:53
mgedmin[trunk] Do this22:53
mgedmin[jfroche-branch-X] Do that22:54
ddaamh22:54
mgedminI forgot revision numbers22:54
ddaain launchpad, the unique name of a branch is ~owner/product/name22:54
ddaaI guess we could use just the name part, but then it's not unique for a product...22:55
ignashmm22:56
ddaamaybe just '[%s %s]' % (owner.name, branch.name)22:56
ignasok, sorry but i have to go now :/22:56
ignas11PM here22:56
ddaathanks a lot22:56
ignasthank you :)22:56
* ddaa -> dinner22:57
*** ignas has quit IRC22:57
*** mgedmin has quit IRC23:19
*** jfroche has joined #schooltool23:53

Generated by irclog2html.py 2.15.1 by Marius Gedminas - find it at mg.pov.lt!