/srv/irclogs.ubuntu.com/2008/01/10/#bzr.txt

=== Verterok is now known as Veterok__
fullermdSheesh, it's quiet today.  Did I miss a memo about a picnic?06:09
quik_hey folks06:26
quik_if `bzr break-lock` isn't breaking the lock set yesterday.. what does one do?06:27
Pengquik_: Are you running it on the URL of the locked branch?06:27
quik_eh?06:28
Pengquik_: Are you trying to break the lock on your local branch or the server you push to or what?06:28
quik_the server I want to push to06:28
Pengquik_: Does "bzr break-lock" prompt you about the lock or do nothing?06:28
quik_seemingly, nothing06:29
Pengquik_: You need to pass it the URL of the branch. "bzr break-lock sftp://..." or whatever.06:29
quik_ah right06:30
Peng:)06:31
quik_thanks Peng06:31
PengYou're welcome. :)06:32
=== asac_ is now known as asac
jmhodgeshey, i can't bzr rspush to, for instance, deploy@foobar:~/thing. it tells me its not a branch or empty directory.  however, i have previously bzr pushed to that same directory.  is this a problem with the macports install of bzr and bzrtools or am i missing something?07:50
fullermdYes   :)07:52
fullermdI have no real knowledge of the domain.  But I have a hard time picturing a way the install could be horked up, and it would work well enough to run, but be broken enough to look in the wrong place.07:52
fullermdYou sure you're not accidentally push'ing instead of rspush'ing?07:53
lifelessjmhodges: what does ls -l of that directory show ?08:19
jmhodgeslifeless: not the place for unix quesions. this is #bzr08:20
jmhodgeslifeless: the #linux or #bsd or whatever you're on :)08:20
jmhodgess/the/try/08:20
lifelessjmhodges: also rspush is very old; you may prefer push + jam's post-push rsync plugin08:20
lifelessjmhodges: I'm helping you debug it, but whatever.08:20
jmhodgeslifeless: jesus i'm sorry i misread08:21
lifelessjmhodges: I'm one of the core authors of bzr.08:21
lifelessjmhodges: still, I meant 'ls -al'08:21
jmhodgeslifeless: i thought you asked me what ls -l always does08:21
jmhodgeslifeless: like a newbie asking a question about ls08:21
lifelessheh08:21
jmhodgeslifeless: i was like "wtf? havnt i seen this guy here before?"08:21
jmhodgessorry08:21
jmhodgesone sec08:21
jmhodgeslifeless: its just .bzr08:22
lifelessok08:22
jmhodgeslifeless: from the pushes i did before08:22
lifelesscat .bzr/branch-format, and .bzr/repository/format08:22
jmhodgeslifeless: locally or in the remote?08:22
lifelessin the remote08:22
jmhodgesbranch-format: Bazaar-NG meta directory, format 1   and format: Bazaar pack repository format 1 (needs bzr 0.92)08:23
jmhodgesthat looks.. wrong somehow08:23
jmhodgesbzr --version says "Bazaar (bzr) 1.0.0"08:23
lifelessok, thats all good08:24
jmhodgesyay08:24
lifelessthe most likely thing then is that the path you are giving is not resolving correctly; could you try an absolute path ?08:24
lifelessdeploy@foobar:/path/to/thing08:24
* jmhodges nods08:24
jmhodgesone sec, need to ci some changes08:25
jmhodgesbzr rspush deploy@biggecko:/home/deploy/bzr/atelier-main # => bzr: ERROR: Remote location is not a bzr branch (or empty directory)08:26
jmhodgeslifeless: ^08:26
jmhodgeslifeless: again, sorry for being a jerk earlier.  i blame the couple of hours of sleep i've been getting for the past week and my own inattention.08:37
lifelessinteresting08:37
fullermdDoes rspush stuff anything in .bzr.log?08:37
lifelessforget about it08:37
lifelessjmhodges: I really recommend you try jam's plugin instead of rspush08:38
lifelessjmhodges: it ssh's in and runs bzr update08:38
jmhodgeslifeless: jams plugin?08:38
jmhodgesi've been out of bzr for too long08:38
* jmhodges googles08:38
fullermdpush-and-update, I think.08:38
fullermdSomething like that.08:38
jmhodgesfullermd: that's it08:39
jmhodgescool.  i'm assuming then that speed gains of rspush aren't worth other problems that i dont know about (other than "it not working")?08:40
lifelessoh, rspush is likely slower08:40
fullermdActually, with packs, the speed gains of rspush can become very negative.08:40
jmhodgesohhh08:40
jmhodgesthanks!08:40
jmhodgesi'll just grab it out of trunk then.  thanks again08:41
jmhodgeshm.. it pushes fine, and the command shows up in `help commands` but bzr push doesn't "do what i expected" or does it really require running `bzr push-and-update` ?08:47
jmhodgesweird08:48
fullermdProbably, if that shows up in the command list.08:48
lifelesstry bzr help push-and-update :)08:49
jmhodgesnatch.  for some reason, i thought it "overwrote" push08:54
=== Laurens is now known as LaurensS
Laurens_SHello, is there someone here who has experience using bzr 1.0, with bzr+ssh on windows? I can't seem to get it to work. sftp works fine, but I'd like the speedimprovement of bzr+ssh which I've seen in linux09:23
spivLaurens_S: what problem are you having?09:30
Laurens_Sthis is the error I get:09:31
Laurens_S(reproducing error.....)09:32
Laurens_SZ:\src>bzr log bzr+ssh://username@server/full/path09:34
Laurens_SConnected (version 2.0, client OpenSSH_3.8.1p1)09:34
Laurens_SSSH username@server password:09:34
Laurens_SAuthentication (password) successful!09:34
Laurens_SSecsh channel 1 opened.09:34
Laurens_Sbzr: ERROR: Connection closed: please check connectivity and permissions (and try -Dhpss if further diagnosis is required)09:34
Laurens_SZ:\src\>09:34
Laurens_SI tried -Dhpss, didn't show anything more09:34
spivLaurens_S: -Dhpss puts more information in the .bzr.log file09:34
Laurens_Sah :)09:35
Laurens_SI'll try again09:35
spiv("bzr version" should report the location of that file)09:35
spivSo, put the contents of the log file on pastebin.com09:36
Laurens_Sah, I finally figured it out:09:37
spivI'm guessing it's something wrong on the server side, perhaps "bzr" is not on the PATH?09:37
Laurens_Syes, that was indeed the case09:37
spivHeh, lucky guess :)09:37
Laurens_SI set BZR_REMOTE_PATH to the correct location now09:37
Laurens_SI thought I did that, but SET revealed that I messed up :09:38
spivAh, right.09:38
Laurens_Szbr_remote_path=/home/ls/local/bin/bzr is not really the same as BZR_REMOTE_PATH :)09:38
spivheh.09:38
Laurens_Swell, thanks for helping out :)09:38
spivYou're welcome :)09:38
Laurens_Snow it turns out I forgot to do a bzr bind, bzr commit before leaving the house :)09:39
Laurens_Sanyway, thanks again, and have a good day y'all :) Loving bzr!09:39
Kinnisonlifeless: Has anyone come up with the answer for you for the traffic shaping?09:57
Kinnisonlifeless: If not, my recommendation is to use 'tc' but it's arcane and confusing, even for people who already know how to use it09:57
gusI broke my branch :(10:02
gusanyone know what this means:10:02
AfCTry using gun tape. That'll fix anything10:02
gus% bzr commit10:02
gusbzr: ERROR: Cannot commit to branch BzrBranch6('file:///home/gus/debian/libapache-sessionx-perl/debian/'). It is bound to BzrBranch6('sftp://mongrel/home/gus/archives/libapache-sessionx-perl/debian/'), which is bound to sftp://mongrel/home/gus/archives/libapache-sessionx-perl/debian/.10:02
gus 10:02
gusits not real informative10:03
AfCgus: Sounds like you don't have write permissions on the target branch, but I assume this used to work for you?10:03
gusyes10:03
AfCgus: anyone else committing to this branch?10:04
gusit was a lightweight checkout, but I removed that and did a heavyweight checkout10:04
gusnope, its all me10:04
fullermdEr.  It's recursive.10:04
AfCgus:  (ie, beware debian style group-per-user nonsense which can result in various users trampling permissions in a repo)10:04
gusafc: its my laptop and home desktop - i don't think there are other users here10:04
fullermdYou've got a branch bound to itself.  That kinda doesn't work...10:05
AfCgus: and as another aside, any particular reason you're using a bound branch ("checkout") instead of a full power branch?10:05
gusI don't know, I'm not really a power bzr user10:05
Kinnisongus: log onto mongrel, cd into /home/gus/archives/libapache-sessionx-perl/debian/10:05
Kinnisongus: type bzr unbind10:05
Kinnisongus then try again10:05
Kinnisongus: As fullermd says, your remote branch is bound to itself, which is the complaint10:06
gusworks fine10:06
gusinteresting10:06
gusbefore the unbind, it said this:10:06
gus% bzr info10:06
gusRepository bound branch (format: dirstate-tags)10:06
gusLocation:10:06
gus  shared repository: /home/gus/archives/libapache-sessionx-perl10:06
gus  repository branch: .10:06
gus 10:06
gusRelated branches:10:06
gus    push branch:10:06
gus  parent branch: /home/gus/archives/libapache-sessionx-perl/upstream10:06
gus 10:06
guswhere's the bit that tells me where its bound to?10:07
Kinnisongus: oddness, it doesn't specifically say it was10:07
Kinnisonit should say checkout of branch: .....10:08
Kinnisonbut then again, it was bound to itself, so who know how odd it could have gotten10:08
fullermdMy checkouts say "Repository checkout (format: dirstate-tags)"10:08
Kinnisongus: fyi, in future, please use a pastebin like pastebin.ca or rafb.net/paste/ for anything more than a line or two10:08
gusmeh. I haven't touched these for ages, perhaps I typed something dumb once ages ago10:08
guskinnison: yep, sorry I'm being lazy.  me--10:09
fullermdWhat version of bzr are you running?10:09
Kinnisonfullermd: oh true, repos will say that10:09
Kinnisongus: and it does say "Repository bound branch" at the top10:09
gusI now have bzr 1.0 at both ends, but these repositories were created / last-modified by a much older bzr version (I couldn't tell you how old)10:09
gusanyway, thanks.  I can pretend I'm still an active debian developer now.  ;)10:10
KinnisonIf you have 1.0, consider using pack format for faster operation (make backups first obv.)10:12
KinnisonI've found pack to be much faster when working with remotely bound branches10:12
KinnisonBut then again, I am odd :-)10:12
gusheh.10:13
gusas an aside: I had lots of trouble finding a way to convert a lightweight checkout into a heavyweight.  Is there some way of doing that 'in place'?10:13
fullermdI think reconfig can do that.10:14
fullermdAt least in recent versions.10:14
gusdidn't seem to (at least in 1.0)10:14
=== cprov-out is now known as cprov
fullermdIt's got --checkout in 1.0...10:16
gusright, but both heavyweight and lightweight checkouts are checkouts, right?10:17
gusor have I confused some terminology somewhere10:17
gusI guess I want a bound branch or something10:17
fullermdUnqualified 'checkout' implies heavyweight.10:17
gusah ok.  I'll give that a go next time.10:18
i386lifeless: ping10:54
=== asac_ is now known as asac
dgl_hi, where can I find something really useful about bzr? I guess documentation at http://bazaar-vcs.org are not very clear.12:21
datodgl_: try http://doc.bazaar-vcs.org/12:21
dgl_dato: I've already been there, sorry about that but I really don't think bzr has a complete documentation that can explain it at all .12:24
datoexplain it all in what sense?12:24
datowhat all? using should be well covered by http://doc.bazaar-vcs.org/latest/en/user-guide/index.html12:25
datousage12:25
dgl_dato: I've been svn user since a couple of years ago and it has a clear workflow description. I know bzr has a lot of possible workflows, but none of them are really explained.12:27
datonot really explained *where*? I haven't read the user guide myself, but the table of contents *hints* like as if they are explained there12:28
dgl_dato: I've read it, it just superficially describes some workflows.12:35
luksanything in particular that you think is missing there?12:36
dgl_dato: I guess, as svn user, learn decentralized workflows is a little hard.12:39
datohttp://doc.bazaar-vcs.org/latest/en/user-guide/index.html#team-collaboration-distributed-style has the very basics, but honestly, there's not *much* more12:40
dgl_dato: and there are some concepts that works little different, like branches, where in svn is just a copy12:41
lukshttp://doc.bazaar-vcs.org/latest/en/user-guide/index.html#branch12:41
datodgl_: http://doc.bazaar-vcs.org/latest/en/user-reference/bzr_man.html#concepts12:41
datooh, what luks says is better12:42
datobut using DAG in that section, eek12:42
* fullermd finds copies a buttload harder to understand.12:43
dgl_gutsy uses bzr 0.9 and my local doc are a little bit smaller12:45
fullermd0.9?!12:46
dato0.9012:46
datodgl_: the docs got *really* polished for 1.012:46
fullermdJeez.  Don't DO that to my poor brain...12:46
dgl_sorry 0.9012:47
datodgl_: if you could read or download the online versions, I think it'd be very helpful12:47
datodgl_: most of it if not all will apply12:47
dgl_dato: I'll read online version12:48
datogreat12:48
dgl_dato: maybe I'll do some pinning to get 1.012:50
datothere ought to be backports for gutsy in bazaar-vcs.org/releases/debs12:51
zurguttHi folks.  Is there gui for bzr, or some ide/editor with integration?13:11
dgl_olive13:13
dgl_zurgutt: olive is a gui for bzr13:14
zurguttthanks dgl13:16
smagounWe have a bzr repository on a server. I'm interested in running14:41
smagouna script whenever someone pushes to that repo. Is there an easy way to do that?14:42
smagoun(I want to run the script on the server side, and ideally don't want to install a plugin on every developer workstation)14:42
mwh_well14:43
mwh_if you're pushing over a dumb transport that sounds kinda hard14:44
smagounmwh_: I was afraid someone would say that. Polling the repo is the obvious solution then?14:44
mwh_if it's bzr+ssh only then something might be possible14:45
mwh_but, hm, probably not easy14:45
=== asak_ is now known as asak
ndimMeh. Google for "git to bzr conversion" and the first hit is titled "How To Convert Bazaar Repo to Git".15:31
LeoNerdUse Tailor15:31
ndimLeoNerd: Is http://bazaar-vcs.org/BzrForeignBranches/Git known to be broken?15:32
ndimI'm looking for a one-off conversion. I have a few git branches based on old CVS versions of Mailman, and Mailman has since switched to bzr.15:33
LeoNerdDunno.. never seen it before. Try it :)15:33
ndimGood :)15:33
LeoNerdIf it's just a oneoff conversion, do you care about history?15:33
LeoNerdIf not, it might be easier just to do an export/import snapshot, and forget history15:33
ndimI care about the stuff I did.15:33
ndimHistory-wise.15:33
ndimIt will probably be easier to merge 50 line patches than one 2000 line patch.15:34
ndimdivide et impera and stuff. :)15:34
luksndim: the converted branch probably won't help you much if you want to merge it to the current mailman bzr branch15:36
luksrevisions will have different IDs, files will have different IDs15:37
ndim$ bzr branch mailman-2.1.7.cg mm-git2bzr-test15:45
ndimbzr: ERROR: exceptions.AttributeError: 'module' object has no attribute '_output_one_line'15:45
ndimluks: Uhm.15:45
lukswhat does ~/.bzr.log say about the error?15:46
ndimSounds like I'm better off converting current mailman from bzr to git, rebase my git stuff on that converted thing, and then convert the merged thing back to bzr...15:46
luksndim: converting from git won't get you branch mergeable with trunk either15:47
ndimluks: traceback in stgit.15:47
luksoh15:47
luksmailman-2.1.7.cg is a git branch?15:47
ndimyupp.15:48
ndimcogito, to be exact.15:48
ndimbzr-git/stgit incompatibility.15:48
ndimThe conversion from git to bzr I could do as generating a few dozen patches and commiting them with bzr.15:49
ndimThat result would then be mergeable.15:49
ndimAs I'd start on the bzr side with real upstream's bzr repo.15:49
luksah15:49
=== Veterok__ is now known as Verterok_
datojelmer: any chance to look at #180128?16:07
datobug #18012816:07
ubotuLaunchpad bug 180128 in bzr-svn "Cannot push to a branch `svn cp'ied`" [Undecided,New] https://launchpad.net/bugs/18012816:07
jelmerwill have a look in the next few days16:07
datook, no hurries, just thought I'd ask.16:07
jelmerthere's a couple of other bzr-svn issues I have to look at as well, and I prefer doing them in batches16:07
datovery well16:07
* jelmer is in Samba-mode atm :-)16:07
ubotuNew bug: #181811 in bzr-git "bzr-git incompatible with certain stgit versions" [Undecided,New] https://launchpad.net/bugs/18181116:30
jelmerddaa: btw, I made a couple of small fixes to your bzr-git branch16:36
ddaagreat16:36
jelmerhttps://code.edge.launchpad.net/~jelmer/bzr-git/jelmer16:36
jelmerfixes support for python2.416:36
ddaajelmer: I do not plan to put any more work on bzr-git in the near future.16:37
jelmerok16:37
ddaafeel free to own it16:37
jelmerI'm not sure I want to :-)16:37
ddaawell, it actually has the beginning of a useful test suite :)16:37
jelmerI may end up owning it though, if I start using it for Samba16:37
muffinresearchQuick question. Does anyone think it would be useful to have a shell script installer for mac that installs bazaar including all of the dependencies16:48
muffinresearchDoes anything like that exist already?16:48
jelmermuffinresearch, I know there were some people working on it16:49
jelmermuffinresearch, not sure what the status is though16:49
jelmerphanatic: ping16:49
muffinresearchI see this: http://www.nabble.com/Bazaar-OS-X-Installer-status-td14254776.html16:49
phanaticjelmer: pong16:49
jelmerphanatic: ^^16:50
phanaticmuffinresearch: there is already an installer available for leopard (bazaar 1.0)16:50
phanaticmuffinresearch: http://bazaar-vcs.org/Download16:51
muffinresearchI don't have leopard16:52
muffinresearchPresumably macports is kept fairly up to date.16:53
phanaticindeed16:53
phanaticthat works fine as well16:53
phanaticjelmer: i could even get bzr-gtk working with macports, it's pretty cool :)16:54
muffinresearchThe main reason is just providing a quick way for colleagues to get up and running with the latest version. I normally install from source myself16:54
muffinresearchmacports is probably the best option. I'll get someone to try that16:57
nebuchadnezzarhello17:37
jelmerhi17:38
nebuchadnezzarI'm following the bzr upgrade in my debian SID and I see some new format, rich-root and rich-root-pack, I don't find documentation/specification about them.17:55
=== mw is now known as mw|food
brianpeirishello18:48
brianpeirisquick question, am I correct in assuming that bzr diff does not support UTF-16 ?18:49
jelmerbrianpeiris: AFAIK, yes18:50
jelmerbrianpeiris: you can however use it with external diff tools18:50
luksbut it will print utf-1618:50
brianpeirisok, explains the "binary files ... differ" message I was getting while trying to diff a UTF-16 sql script.18:51
luksactually, it won't print anything18:51
luksyeah18:51
luksbzr checks if the file has nul bytes, and if yes it assumes it's binary18:51
brianpeirisalrighty, thanks for the help.18:54
ubotuNew bug: #181842 in bzr "BZR_SSH environment variable is not explained in docs" [Undecided,New] https://launchpad.net/bugs/18184218:55
dglhi18:55
dglCan anyone explain me what are the differences between init and init-repos?18:56
jelmerinit creates a branch18:57
jelmerinit-repo creates a (shared) repository18:57
dgljelmer: ok, i've already read that, but what does it mean?18:58
dglbasically you create a branch just few times and then start to 'branch' it, isn't it?19:01
jelmerdgl: see http://doc.bazaar-vcs.org/bzr.dev/en/user-guide/index.html#core-concepts for an explanation of branch and repository19:01
brianpeiris@dgl, see also: http://doc.bazaar-vcs.org/bzr.dev/en/user-reference/bzr_man.html#repositories19:04
dglthanks, I will19:05
UbuntuStatsHi - From now on #bzr will have statistics on ubuntuircstats.org (gouki)19:29
dglwhere is a nice place to create a shared repos at an ubuntu server?19:32
goukiHere it is: http://ubuntuircstats.org/bzr.html19:32
goukiEnjoy guys!19:32
=== bigdo1 is now known as bigdog
=== mw|food is now known as mw
ubotuNew bug: #181855 in bzr "cygwin bzr branch crashes with IOError: [Errno 0] Error" [Undecided,New] https://launchpad.net/bugs/18185520:01
dglwhere is a nice place to create a my mainline shared repos at an ubuntu server?20:16
beunodgl, Launchpad es where you create branches, but it doesn't support shared branches as far as I know20:17
beunoI suppose, at the moment, it's meant for local use, to save space20:17
LarstiQdepends on what you mean with shared20:17
* beuno waves at LarstiQ 20:17
LarstiQmultiple people can push to branches on launchpad, provided they are part of the team owning the branch20:18
LarstiQbeuno: heya :)20:18
dglbeuno: I've got a server with ssh, I just want to use it as a shared repos20:18
LarstiQdgl: you're asking what a good location is to place public repositories?20:18
beunodgl, you want multiple people to be able to work on a branch, or you want multiple branches to use the same repo?20:18
LarstiQin a FHS kinda way20:18
* LarstiQ user /srv/bzr/<repository-name-here>20:19
dglLarstiQ: sorry, but I think I was not clear20:19
dglI was talking about filesystem20:19
beunodgl, that just confuses me more  :D20:20
dglbeuno: LarstiQ's understood me20:21
beunoah, yes, he's very good at that20:21
dglbeuno: I was asking about a nice place to create a shared repos at my server file system20:21
beunodgl, happy you got your answer  :D20:22
LarstiQdgl: it depends on the rest of your filesystem layout of course, /var/www/bzr/ is also not unthinkable :)20:23
dglLarstiQ: and /var/bzr?20:23
LarstiQdgl: not where I'd place it myself20:23
* LarstiQ thinks it fits best under /srv20:23
LarstiQdgl: http://www.pathname.com/fhs/pub/fhs-2.3.html#SRVDATAFORSERVICESPROVIDEDBYSYSTEM20:24
lifeless/srv/bzr/www20:24
lifelesshmm, maybe /bzr/www/srv/var/20:24
LarstiQlifeless: argh :P20:24
lifelessno like?20:25
LarstiQyou'd have to have a rather complex setup for that to make sense I think20:25
lifelessok, ok.20:25
LarstiQsomething like launchpad perhaps20:25
lifeless/var/bzr/srv/www/20:25
LarstiQlifeless: are you pulling my leg again? :)20:25
lifelessmoi?!20:25
dglI just want to use sftp20:25
LarstiQlifeless: true true, I've never caught you doing it before.20:26
lifelessmaybe not on IRC; in person you have:)20:26
LarstiQhah! Now I did ;)20:26
lifelessdgl: anything is fine; key things to think of is 'will it be backed up by your backup programme'20:26
dgllifeless: thanks20:28
* lamont looks around for a bzr god21:12
lamontlifeless: you awake?  and are you a god? :-)21:12
lamontif a user has read access to .bzr/* (and down), but lacks read access on bar/foo.c, how can I tell bzr to just ignore that little issue and keep right on going with its life?21:14
datoheh21:15
=== cprov is now known as cprov-out
lifelesslamont: (True, True)21:15
lamontlifeless: I see that you remember ghostbusters.21:17
lamontnow about bzr and bar/foo.c ....  any hope for me?21:17
lamontlifeless: ^^21:20
lifelesslamont: you can't21:23
jearl\q21:24
lifelesslamont: (in general). Bu perhaps you can give me more info ;)21:24
PengNice, some random doofus clicked on one of my bundles because it's the only Google result for "JGR1my space". :\21:37
PengGooglebot also found a link to /yaygyosr.html in one of them.21:38
lifelessrotfl21:43
PengOoh, the "JGR1my space" search was from Mexico!22:31
lifelessyay hung test :(22:39
LeoNerdHung-like-horse, or hung-like-Sadam ?22:39
PengOuch.22:39
lifelessI'll find out soon22:39
lifelesshung like reading bytes that never came22:40
PengI got a "Connection reset by peer" after 3 minutes pushing 1 KB of changes over bzr+ssh. :D22:40
abentleylifeless: Your patch to remove the need for ParentProvider.get_parents looks very strange.  The goal is fine, but please don't merge in current form.22:55
lifelessabentley: good timing; I had just hit enter to push and submit22:56
lifelesswhats wrong with it ?22:56
abentleyIt means that Graph.parents_provider is sometimes None.22:56
abentleySorry22:56
abentleyGraph.get_parents is sometimes None22:56
abentleyThat's a public API22:56
datoabentley: (when you have a couple minutes and if you don't mind, did you see my question about having send, when -r is specified, always bundle those revisions, even if they're in the submit branch?)22:56
abentleyIt should always be a callable.22:57
lifelessabentley: oh, that wasn't clear to me because its not defined explicitly22:57
abentleyYou can deprecate it, or remove it entirely, but no one expects a method to sometimes be None.22:58
lifelessso if I add22:58
lifeless'if self.get_parents is None: del self.get_parents' you're happy ?22:58
abentleyYou mean Graph.get_parents would sometimes be defined, and sometimes not defined?22:59
lifelessor I can write a regular get_parents method22:59
lifelesswell currently I can supply a ParentsProvider with get_parents = None22:59
abentleyI prefer either deprecate it, or remove it in all cases.22:59
lifelessbecause the Graph class does not define any of what you are referring to22:59
lifelessand we don't provide any thunking in either direction23:00
abentleylifeless: You can, but that's an API violation.23:00
lifelessabentley: how about I write Graph.get_parents and graph.get_parent_map, both in terms of each other23:00
lifelessabentley: then the constructor can replace just one from the ParentsProvider23:01
abentleyWhy would you write Graph.get_parents_map?23:01
abentleyLet's just standardize on ParentsProviders having that.23:01
lifelessso that providers with get_parents still work - that is the point of deprecating that isn't it ?23:01
abentleyThere are no providers with only get_parents.23:02
lifelesslet me rephrase; we appear to have a deprecation which is not really a deprecation - its not compatible. I'm happy to either turn it into a full break; or a real deprecation where the old code works. Your choice.23:02
lifelessabentley: third party repositories will almost certainly be providers with only get_parents23:02
lifelesswhich is public in that Graph.__init__ takes a public parameter.23:02
abentleyI have a slight preference for real deprecation, but I'd be fine with either.23:04
lifelessok I'll whip it up23:04
abentleydato: That has already proven to result in horrible messes, and that's why there's a submit branch.23:05
abentleyHowever, if you want an override, that's fine.23:05
datooverride == option?23:05
abentleyYes.  People have asked for that in the past, but they've never suggested how the commandline would work.23:06
datook, I'll make a note, and possibly talk to you about it in the (near) future. now it'd be bed time for me.23:06
abentleyOne way would be a second revision spec.23:06
abentleyAnother would be an option that says --use-revision-spec-instead-of-submit-branch23:06
dato(I guess I didn't realize the time it was when I brought up the subject, sorry. chao)23:08
abentleyBut remember that a merge directive, is meant to convey the changes you want made.  That's what the -r controls.23:09
abentleyThe revisions supplied are the ones necessary for other people to apply the those changes.23:10
abentleyFor a cherry-pick, that is typically *more* revisions than the -r range includes.23:10
abentleyThe submit branch should indicate which revisions everyone already has.23:11
=== _[PUPPETS]Gonzo is now known as [PUPPETS]Gonzo

Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!