2010-05-06
Next Session: TBD |
1 [07:00] <dholbach> hello everybody!
2 [07:00] <dholbach> who do we have here for the Q&A session?
3 [07:00] <dholbach> maybe let's start with a quick round of introductions
4 === ChanServ changed the topic of #ubuntu-classroom to: Welcome to the Ubuntu Classroom - http://wiki.ubuntu.com/Classroom || Support in #ubuntu || Upcoming Schedule: http://is.gd/8rtIi || Event: Packaging Training Session - Current Session: Packaging and MOTU Q&A - Instructor: dholbach || Questions in #ubuntu-classroom-chat
5 [07:01] * dholbach is Daniel Holbach, living in Berlin, interested in making Ubuntu Development more exciting :)
6 [07:01] <dholbach> who else do we have here?
7 [07:01] <gotunandan> i am here
8 [07:01] <dholbach> sorry
9 [07:01] <dholbach> just turned the moderation off
10 [07:02] <dholbach> so please speak up, when you're here and let's do some introductions :)
11 [07:02] <gotunandan> nobody else ?
12 [07:02] <gotunandan> i saw your dent on identi.ca and popped up here !
13 [07:02] <dholbach> nice gotunandan :)
14 [07:03] <dholbach> I'm sure we have a few shy lurkers among the 154 people in here :)
15 [07:03] <jk-> ok, Jeremy Kerr here, living in Perth, Australia
16 [07:03] <gotunandan> still need to use #ubuntu-classroom-chat for the questions ?
17 [07:03] <jk-> :)
18 [07:03] <dholbach> hey jk-
19 [07:03] <jk-> heya dholbach
20 [07:03] <dholbach> no no, we'll just chat in here
21 [07:04] <dholbach> are you excited about maverick? :)
22 [07:04] <jk-> i am in a constant state of excitement
23 [07:04] <gotunandan> yeap... and more keen on joining in the "fun" in some way
24 [07:05] <dholbach> awesome - I set up a vm for maverick already :)
25 [07:05] <dholbach> did you guys bring any questions or is there anything you were wondering about packaging, ubuntu development, motu, tools, processes, etc?
26 [07:06] <jk-> kinda - I see that a package has been dropped (due to FTBFS), and would like to adopt it
27 [07:06] <jk-> so maybe that's more of a 'process' type issue
28 [07:06] <gotunandan> <QUESTION> What would I need to start with to help with packaging ? I have already done some basic packaging and I have a ppa on launchpad as gotunandan. But not sure whom to approach to get actively involved in helping the MOTU
29 [07:06] <maco> dholbach: better a vm than your actual system O_O toolchain freeze scary
30 [07:07] <dholbach> ok, let's go through these one by one
31 [07:07] <dholbach> jk-: cool - which package is it?
32 [07:07] <jk-> python-creoleparser
33 [07:07] <dholbach> maco: definitely - https://wiki.ubuntu.com/UbuntuDevelopment/UsingDevelopmentReleases is your friend :)
34 [07:08] <dholbach> jk-: I found it here: https://launchpad.net/ubuntu/+source/creoleparser/0.6.1-1
35 [07:09] <dholbach> and it seems to have built on i386: packages that are architecture-independent, such as a bunch of python scripts are just built on i386
36 [07:09] <maco> gotunandan: have you seen harvest http://daniel.holba.ch/harvest ? its dholbach's "find low hanging fruit for motu-wannabes" site
37 [07:10] <dholbach> maco: this one is not functional right now - I hope to get the new harvest site up soon though
38 [07:10] <gotunandan> maco: I have heard of it, not seen it. Having a look
39 [07:10] <jk-> dholbach: but not built for lucid?
40 [07:10] <maco> dholbach: oh
41 [07:10] <dholbach> Usually the https://launchpad.net/ubuntu/+source/creoleparser/+publishinghistory page shows if it was removed or something
42 [07:10] <maco> dholbach: what happened to the prettiness rockstar made last UDS? it never got uploaded?
43 [07:10] <dholbach> jk-: it was built in karmic, then it never changed afterwards
44 [07:11] * maco tries to remember +publishinghistory
45 [07:11] <jk-> i see
46 [07:11] <dholbach> maco: it was improved, but it's not deployed yet
47 [07:11] <dholbach> jk-: we don't rebuild all packages for a new release - it'd probably take weeks :)
48 [07:12] <dholbach> we just "copy them over"
49 [07:12] <dholbach> jk-: is there anything wrong with the package right now?
50 [07:12] <jk-> ok
51 [07:12] * jk- wonders why it's not appearing in the repo then
52 [07:12] <dholbach> ok, let's see about that
53 [07:12] <jk-> possibly quite off-topic now though :)
54 [07:12] <dholbach> no no, that's fine
55 [07:13] <dholbach> first of all: you're right :)
56 [07:13] <persia> It's a bit off-topic, but the binary was probably removed due to a FTBFS with current tools, as a result of a supportable-binaries spec in lucid. This affected ~600 packages.
57 [07:13] <dholbach> daniel@miyazaki:~$ LC_ALL=C sudo apt-get install -s python-creoleparser
58 [07:13] <dholbach> Reading package lists... Done
59 [07:13] <dholbach> Building dependency tree
60 [07:13] <dholbach> Reading state information... Done
61 [07:13] <dholbach> elky: Couldn't find package python-creoleparser
62 [07:13] <dholbach> daniel@miyazaki:~$
63 [07:13] <dholbach> oh, that's interesting, let me test-build it
64 [07:13] <persia> dholbach: Fix your nick-completion :)
65 [07:13] * elky tilts her head wondering...
66 [07:14] <dholbach> errrrrr, that was xchat-gnome-auto-completion
67 [07:14] <dholbach> sorry about that - I didn't even press "tab" :)
68 [07:14] <dholbach> I just copied/pasted
69 [07:14] <elky> That's not a very nice feature then
70 [07:14] <dholbach> not, not really
71 [07:14] <ToyKeeper> xchat -> settings -> preferences -> input box -> [ ] Automatic nick completion (without tab key)
72 [07:15] <persia> It's more helpful than it ought be: file a bug :)
73 [07:15] <dholbach> ok, so I just apt-get source creoleparser; sudo pbuilder build creoleparser*.dsc
74 [07:15] * elky hugs regular xchat and goes back to what she was doing.
75 [07:16] <gotunandan> maco: yes seen it now. Cannot really fathom where to go though ?
76 [07:16] <dholbach> aha, the package fails to build: http://paste.ubuntu.com/428772/
77 [07:16] <maco> gotunandan: dholbach just said that its currently in ick mode. the shiny new harvest isnt online yet apparently
78 [07:16] <gotunandan> maco: the sourcepackages,html page is huge !
79 [07:17] <maco> yeah.... i remember that
80 [07:17] <dholbach> gotunandan: best to ignore that page for now, https://wiki.ubuntu.com/MOTU/TODO lists a bunch of pages that make more sense right now
81 [07:17] <dholbach> jk-: I just had a look if the package changed in Debian and it did - http://packages.debian.org/changelogs/pool/main/c/creoleparser/creoleparser_0.7.1-1/changelog looks a lot like it's going to build now
82 [07:17] <jk-> dholbach: so, say I hack on it a bit and fix that;
83 [07:17] <dholbach> " * Use python.mk for install options."
84 [07:17] <ToyKeeper> (if xchat-gnome doesn't have the UI, it's probably '/set completion_auto off')
85 [07:18] <maco> gotunandan: also, if you can read a bit of code, know how to apply a patch & rebuild a package... the patch review team could use some help ;-)
86 [07:18] <dholbach> I'll just get the source from Debian and see if that works
87 [07:18] <dholbach> (got to the changelog page via packages.debian.org/src:<source package>)
88 [07:18] <maco> gotunandan: the patch review team and motu both have the goal of getting the patch in & bug fixed. its just that the patch review team tries to get the patches to upstream or debian so then the motu can just update the package to the latest version upstream and it magically is fixed
89 [07:19] <gotunandan> maco: yes, thats certainly possible. Whom do I get in touch with, nigelbabu ?
90 [07:19] <jk-> dholbach: sid ?
91 [07:19] <dholbach> jk-: if the package from debian builds just fine, we'll auto-import it in a few days (once maverick is truly open) since the package is unmodified in Ubuntu
92 [07:19] <dholbach> jk-: http://packages.debian.org/search?searchon=sourcenames&keywords=creoleparser
93 [07:19] <maco> gotunandan: sure or join #ubuntu-reviews
94 [07:19] <jk-> ok
95 [07:19] <maco> gotunandan: we've got a patch day going right now
96 [07:19] <persia> gotunandan: If you're looking for patch review, stop by #ubuntu-reviews : if you're looking at getting the patch straight into Ubuntu, #ubuntu-motu is the better place.
97 [07:20] <jk-> but we sync from sid or squeeze?
98 [07:20] <maco> persia: well once its reviewed it can still have a shot at an sru
99 [07:20] <dholbach> jk-: ah, sorry - I misunderstood - I think for maverick we'll sync from sid again
100 [07:20] <jk-> i guess it doesn't matter in this case :)
101 [07:20] <dholbach> yep, but it fails the same way
102 [07:21] <dholbach> http://paste.ubuntu.com/428774/
103 [07:21] <jk-> so I should hack on the debian package instead?
104 [07:21] <dholbach> it'll get auto-imported, so it will make sense to work on that
105 [07:22] <dholbach> what I'd probably try next is to see if a bug was already filed upstream about this
106 [07:22] <dholbach> it's a bit weird that the package seems to build in debian, but not here
107 [07:22] <dholbach> since we don't modify its code
108 [07:23] <dholbach> jk-: if the fix is really just "add some __init__.py files", it'd probably make sense to fix it in lucid too if the package is truly helpful
109 [07:24] <persia> Actually, arch:all packages often run into this issue, because of differences in the Debian and Ubuntu build systems.
110 [07:24] <dholbach> gotunandan: to get back to your question: https://wiki.ubuntu.com/MOTU/TODO lists a lot of useful stuff - once maverick opens I guess a lot of people will focus on merging packages from Debian
111 [07:24] <persia> Most of those are bugs that are happily fixed in Debian.
112 [07:24] <dholbach> persia: but you wouldn't know what specifically is broken there? it's not a common ftbfs?
113 [07:25] <persia> I don't happen to know, sorry.
114 [07:25] <dholbach> ok, thanks
115 [07:25] <dholbach> gotunandan: https://wiki.ubuntu.com/UbuntuDevelopment/Merging should give you some tips how to do it
116 [07:25] <dholbach> (and of course there's always #ubuntu-packaging and #ubuntu-motu)
117 [07:25] <persia> http://qa.ubuntuwire.com/uehs/no_updated.html is also worth mentioning in the context of merging: we don't get everything from Debian.
118 [07:25] <gotunandan> dholbach: thanks !
119 [07:25] <jk-> ok, say it's an ubuntu-specific fix then, and I create a dpatch, what next?
120 [07:26] <dholbach> get it sponsored: https://wiki.ubuntu.com/SponsorshipProcess
121 [07:26] <jk-> ok, cool
122 [07:26] <dholbach> usually just file a bug report or use an existing one, add the patch, subscribe ubuntu-sponsors, get it uploaded
123 [07:27] <maco> dpatch?
124 [07:27] <dholbach> or you can use the distributed development approach
125 [07:27] <dholbach> where you branch off of lp:ubuntu/creoleparser
126 [07:27] <maco> is dpatch like debdiff? ive not heard this one before...
127 [07:27] <jk-> so attach the dpatch itself, or a patch to add the dpatch + changelog + etc
128 [07:27] <dholbach> maco: dpatch is one of the many patch systems
129 [07:27] <jk-> ?
130 [07:27] <persia> maco: dpatch is one of hte available patch systems: `apt-cache show dpatch`
131 [07:27] <dholbach> jk-: the latter makes it more straight-forward to just upload it (if it's good)
132 [07:28] <jk-> ok, great
133 [07:28] <maco> jk-: for what dholbach just mentioned with distributed development: https://wiki.ubuntu.com/DistributedDevelopment/Documentation
134 [07:28] <jk-> maco: thanks
135 [07:28] <dholbach> where you branch off of lp:ubuntu/creoleparser , create a branch for that fix, and add a merge proposal
136 [07:28] <maco> im starting to appreciate source version 3 uniting on quilt. at least i only have to know how one thing works
137 [07:29] <dholbach> mvo wrote a tool call edit-patch (in ubuntu-dev-tools), it should help with all the different patch systems
138 [07:29] <maco> dholbach: hey i have a motu sponsory question. is there a way to see all bugs that have branches attached?
139 [07:29] <maco> or to see all open merge proposals for which you are able to approve it?
140 [07:29] <dholbach> maco: I use http://qa.ubuntu.com/reports/sponsoring/index.html
141 [07:29] <dholbach> and sort by the first column
142 [07:29] <dholbach> all "unseeded"
143 [07:30] <maco> dholbach: i thought that was just listing bugs that have sponsors subscribed
144 [07:30] <dholbach> erm
145 [07:30] <dholbach> that's broken
146 [07:30] <dholbach> maco: no, branches too
147 [07:30] <dholbach> but the "unseeded" thing is broken right now
148 [07:30] <dholbach> more TODO list
149 [07:30] <dholbach> I'll get to it later today
150 [07:30] <maco> it just looks like all but 2 are for unseeded packages
151 [07:31] <jk-> does the merge proposal need a corresponding bug?
152 [07:31] <dholbach> maco: yes, that's a bug
153 [07:31] <maco> jk-: generally, yeah
154 [07:31] <dholbach> no
155 [07:31] <dholbach> if there's a bug, link to it, but you don't need to specifically file onw
156 [07:31] <dholbach> one
157 [07:31] <maco> i thought for bookkeeping it was good to have a bug and a LP: #foo in the changelog?
158 [07:31] <jk-> ok
159 [07:32] <maco> is this a bzr-lets-bookkeeping-happen-regardless thing?
160 [07:32] <dholbach> if there's some actual discussion about the thing going on, it's good to link to it in the changelog
161 [07:32] <dholbach> but if you just say something like "changed x, y and z in a, b, c: this makes the package build again" that's totally fine
162 [07:32] <persia> If there is a bug, the bug in the changelog is good. For a Debian merge or UEHS update, no point to a bug if it doesn't already exist and one is pushing a branch.
163 [07:33] <dholbach> you'd write the same in the bug report anyway
164 [07:34] <maco> persia: k so the tendency i had to only file a bug once i had a patch ready to hand to a sponsor... that would no longer require filing a bug?
165 [07:34] <dholbach> not if you use distributed development
166 [07:34] <persia> maco: You may either file a bug or push a branch and request a merge. Both show up on the sponsoring report.
167 [07:34] <dholbach> but still document as good as you can :)
168 [07:34] <maco> neat
169 [07:34] <dholbach> when you use bzr and distributed development, you can easily do a bzr commit --fixes lp:123456 to link it
170 [07:35] <maco> though rawr ive only been at this a year-ish and im already behind the times
171 [07:35] <dholbach> (if this bug exists)
172 [07:35] * persia tends to put the bugs in the changelog, and then use debcommit to automate the --fixes bit
173 [07:35] <dholbach> maco: I think everybody's figuring out how best to make use of UDD (ubuntu distributed development) right now
174 [07:35] * maco does like persia
175 [07:35] <dholbach> persia: yes, that works as well :)
176 [07:35] <jk-> dholbach, persia, maco: awesome, thanks for the info, i owe you guys a beer :)
177 [07:35] <dholbach> (I've been working a lot on non-distro things in the last few weeks :-))
178 [07:36] <dholbach> jk-: no worries
179 [07:36] <persia> If it's not clear to anyone: there are almost no "right answers", as long as you find something that 1) works for you, and 2) complies with policy, and 3) happens to match others workflows.
180 [07:36] <maco> persia: you know whats fun (not really)? this mess: bzr-buildpackage -S -- -sa -v####
181 [07:36] <dholbach> as you can see it was very useful to talk about these questions :)
182 [07:36] <dholbach> maco: I'd move the "-S" behind the "--"
183 [07:37] <dholbach> bzr bd -- -S -sa -v<version>
184 [07:37] <persia> Why -sa?
185 [07:37] <maco> persia: i was merging fromd debian
186 [07:37] * persia actively avoids using -sa *except* when cleaning up after other folks
187 [07:37] <dholbach> persia: if you need to upload a new upstream tarball?
188 [07:37] <maco> where there was a new upstream release
189 [07:37] <persia> Still shouldn't need -sa
190 [07:37] <maco> well buildpackage was Not Happy when i didnt use it
191 [07:37] <maco> though that couldve been because i forgot -v too
192 [07:37] <persia> dholbach: It will auto-include it if it thinks it's warranted.
193 [07:38] <dholbach> persia: if you have 1.2-1ubuntu1 in Ubuntu and merge 2.4-5 from Debian?
194 [07:38] <dholbach> oh nice
195 [07:38] <maco> i was flailing about today trying to remember how to merge
196 [07:38] <persia> RIght, that's clearly a new upstream, so you get a new tarball.
197 [07:38] <dholbach> persia: and it includes the -sa magically?
198 [07:38] <maco> persia: so my forgetting -v is why it wanted me to -sa?
199 [07:38] <persia> It only fails when the versions aren't sufficiently different, or you need to push a new tarball to clean up after some mess.
200 [07:38] <dholbach> maco: the two are unrelated
201 [07:38] <maco> hrmph
202 [07:38] <persia> maco: Yes, because you didn't scope it correctly.
203 [07:38] <dholbach> gotcha
204 [07:39] <dholbach> do you all have any more questions? anything that was unclear in the explanations right now?
205 [07:39] * maco looks back and forth from persia to dholbach
206 [07:39] <persia> dholbach: If the upload version and the prior version have different upsteam versions, yep.
207 [07:39] * maco wishes you'd agree a bit
208 [07:39] <dholbach> maybe somebody of the shy people (not jk- and gotunandan :-)) who are just lurking here?
209 [07:39] <dholbach> persia: great
210 [07:39] <dholbach> maco: I filed bug 576242
211 [07:39] <ubottu> Launchpad bug 576242 in ubuntu-sponsoring "Everything is "unseeded"" [Undecided,New] https://launchpad.net/bugs/576242
212 [07:39] <persia> maco: It is the dynamic cooperative disagreement that is central to the process of arriving towards a flexible set of working practices :)
213 [07:40] <maco> dholbach: it *is* possible that one of those very rare times when the main sponsors are more efficient than the universe sponsors happened
214 [07:40] <maco> (yes i know old names)
215 [07:40] <dholbach> maco: no, I think it's more a problem with the packagesets in maverick
216 [07:40] <persia> It's a bug.
217 [07:41] <dholbach> any more questions? anything else you wondered about? anything else that seems broken? :)
218 [07:42] <maco> how many uploads do you have to sponsor before you get a pony?
219 [07:42] <maco> </jorge>
220 [07:43] <dholbach> I don't know, but you get a lot of people liking you for reviewing their stuff :)
221 [07:46] <dholbach> any more questions?
222 [07:46] <ajmitch> how do we be a rock star? :)
223 [07:47] <dholbach> ajmitch: you of all should know!
224 [07:47] <arand_> (Don't have a system in fron of me, so general question): I find that I struggle quite a bit when i try to update a package with new [insert scm] versions. If I don't have an orig.tar.gz, is there any general guidelines on how to get to the point where ther is one, whilst still reusing the old diff.gz (possible?)
225 [07:47] <maco> scm?
226 [07:47] <ajmitch> source control
227 [07:47] <arand_> git/svn/whathaveyou
228 [07:48] <maco> pristine-tar?
229 [07:48] <ajmitch> generally there'll be ways of exporting it, such as 'svn export', or possibly 'make dist' for an automake project
230 [07:49] <ajmitch> whatevr way suits the project best, you'll want to have a get-orig-source target in debian/rules so that it can be automated by others
231 [07:49] <dholbach> ./configure <---> && make dist (if available) should give you a tarball
232 [07:49] <dholbach> and something like svn export should make sure you don't have revision control meta data in your tarball
233 [07:50] <dholbach> I hope that one fine day in the future we stop caring about tarballs and just use branches for everything :)
234 [07:50] <maco> what does "bzr-buildpackage --split" do?
235 [07:50] <dholbach> until then, the above will make sense :)
236 [07:50] <maco> it says Automatically create an .orig.tar.gz from a full source branch.
237 [07:50] <maco> dholbach: does that do what arand_ is asking for?
238 [07:50] <ClassBot> There are are 10 minutes remaining in the current session.
239 [07:50] <arand_> The autogen.sh I've seen around (with no configure), should one just run that and then "make dist" In those cases?
240 [07:51] <persia> arand_: All those alternatives aside, you should encourage any upsteam to produce some release archive (.tgz, .tar.bz2, .zip, whatever) when they do a release.
241 [07:51] <dholbach> arand_: that sounds like a good bet
242 [07:51] <maco> arand_: autogen.sh, in my experience, generates the configure
243 [07:51] <dholbach> maco: yes, that should work if they use bzr (or one has set up a launchpad import of some other revision control branch)
244 [07:52] <arand_> Does the autogen not also run the configure normally?
245 [07:52] <maco> dholbach: idea for next UOW: UDD
246 [07:52] <ajmitch> yes please
247 [07:52] <dholbach> arand_: I've seen it do so, but I wouldn't rely on it
248 [07:52] <dholbach> arand_: best read what it does - they're usually quite small
249 [07:52] <dholbach> maco: james_w has been running them for every UDW and a few UOWs too
250 === gotunandan_ is now known as gotunandan
251 [07:53] <dholbach> ajmitch: there should be heaps of logs :)
252 [07:53] <arand_> Anyway, it appears as though "make dist" was the link I was missing in general
253 [07:53] <arand_> Thanks
254 [07:53] <dholbach> glad that works :)
255 [07:53] <maco> dholbach: oh in that case i need to pay more attention :P
256 [07:53] <dholbach> maco: but yeah, I guess we need to raise the profile of UDD
257 [07:53] <dholbach> I just learned about bzr lp-open a few weeks back
258 [07:53] <ajmitch> dholbach: yeah, it'll be a matter of finding & reading, like much of the documentation around
259 [07:54] <ajmitch> a few weeks? I learnt about it today
260 [07:54] <dholbach> ajmitch: https://wiki.ubuntu.com/UbuntuDeveloperWeek/Previous
261 [07:54] <ajmitch> thanks
262 [07:54] <dholbach> de rien
263 [07:54] <maco> if the goal is for that to be the way things are done moving forward, seems it ought to be whats given to new contribs off the bat
264 [07:55] <imbrandon> bzr lp-open ?
265 [07:55] <dholbach> imbrandon: will open the launchpad page of the branch in a browser :)
266 [07:55] <persia> There's a sufficiency of folks that prefer the non-bzr method that we'll probably end up supporting both for a very long time.
267 [07:55] <dholbach> so if you want to submit a merge proposal, that makes it easy
268 [07:55] <persia> (especially for the die-hard git folk)
269 [07:55] <ClassBot> There are are 5 minutes remaining in the current session.
270 [07:56] <dholbach> ok, any more questions?
271 [07:58] <dholbach> ok... in that case best ask your questions in #ubuntu-packaging or #ubuntu-motu later on!
272 [07:58] <dholbach> have a great day everybody and thanks for all your questions and answers
273 [07:58] <dholbach> you guys rock!
274 [07:58] <ajmitch> thank you dholbach
275 [07:58] <arand_> Thanks for the session!
Packaging/Training/Logs/2010-05-06 (last edited 2010-05-28 11:33:36 by i59F77AA7)