2009-08-13

   1 [13:00] <@dholbach> HELLO EVERYBODY!
   2 [13:00] <@dholbach> This is the first time we break out of the tutorial-esque or Q&A-like way of Packaging Training sessions and decided to have an "On-Call Review" session instead. I'm sure we'll all learn something new today and have some fun.
   3 [13:00] <@dholbach> we have cjwatson, seb128 and james_w here, but I noticed a bunch of others who are likely able to help out a bit with questions :-)
   4 [13:00] <@seb128> hello!
   5 [13:00] <@cjwatson> hiya
   6 [13:00] <@james_w> hi!
   7 [13:00] <@dholbach> who do we have here for the session today and who brought some review requests? :-)
   8 [13:01] <ara> o/
   9 [13:01] <@dholbach> hey ara
  10 [13:01] <@seb128> hey ara ;-)
  11 [13:01] <juanje> hi
  12 [13:01]  * sebner waves
  13 [13:01]  * noodles775 waves too
  14 [13:02] <@dholbach> While we're looking for somebody to pluck up some courage, we could take a look at http://people.canonical.com/~dholbach/sponsoring/ :-)
  15 [13:02] <@cjwatson> well then, why don't we go first-come-first-served
  16 [13:02] <juanje> dholbach: I'm up
  17 [13:02] <@cjwatson> ara spoke first :)
  18 [13:02] <juanje> ;-)
  19 [13:02] <ara> juanje, you go or I go?
  20 [13:03] <juanje> ara: you, please
  21 [13:03] <ara> ok, bug https://bugs.launchpad.net/ubuntu/+source/ldtp/+bug/412882
  22 [13:03] <ara> new upstream version of ldtp
  23 [13:03] <@dholbach> I guess we could review packages simultaneously and just say [package A] at the beginning of our sentences to make the logs make more sense afterwards?
  24 [13:04] <ara> [package ldtp] OK ;-)
  25 [13:04] <@cjwatson> [ldtp] ok, an upstream merge, so I'll start by fetching the current package from Ubuntu and the Debian package
  26 [13:04] <ara> [package ldtp] yes, cjwatson, that's what I did
  27 [13:04] <@dholbach> juanje: what is your review request?
  28 [13:04] <@cjwatson> [ldtp] I have a chdist setup to make it trivial to fetch Debian versions of things with 'chdist apt-get unstable -d source ldtp'
  29 [13:05] <ara> [package ldtp] but I also wanted to fix bug https://bugs.launchpad.net/ubuntu/+source/ldtp/+bug/412871
  30 [13:05] <juanje> [nautilus-md5sum] For example -> http://revu.ubuntuwire.com/p/nautilus-md5sum
  31 [13:05]  * dholbach makes a note to investigate chdist afterwards. :-)
  32 [13:05] <@dholbach> [nautilus-md5sum] getting the source code with dget -xu
  33 [13:05] <@cjwatson> [ldtp] ara: have you contacted the Debian maintainer of this package?
  34 [13:05] <ara> [ldtp] finally I linked the upstream version, attached the diff.gz and subscribe universe sponsors
  35 [13:05] <dutchie> hmm, how long does it take an upload to show up on revu?
  36 [13:06] <ara> [ldtp] cjwatson, yes, we communicate quite often
  37 [13:06] <@seb128> [nautilus-md5sum] notice the copyright warning on the page
  38 [13:06] <@cjwatson> [ldtp] with upstream updates it's important that we both end up using the same .orig.tar.gz, otherwise it's hard to sync later. ok.
  39 [13:06] <@seb128> [nautilus-md5sum]  that's a blocker to get new packages accepted
  40 [13:06] <@dholbach> dutchie: can you ask on #ubuntu-motu if there's some REVU admins about who can answer the question?
  41 [13:06] <dutchie> will do
  42 [13:07] <juanje> [nautilus-md5sum] But I don't see any error, just a legal message, but no error
  43 [13:07] <@dholbach> [nautilus-md5sum] running a test-build in pbuilder while I'm taking a look at everything else. juanje: do you think you could make the version 0.3.1-0ubuntu1 instead to indicate that it is not in Debian yet?
  44 [13:07] <@cjwatson> [ldtp] the interdiff is tiny, and looks mostly correct: http://paste.ubuntu.com/252457/
  45 [13:08] <@cjwatson> [ldtp] you should normally *not* update Standards-Version in Ubuntu changes
  46 [13:08] <@seb128> [nautilus-md5sum] $ licensecheck * -r
  47 [13:08] <@seb128> nautilus-md5sum.py: GPL (v2 or later)
  48 [13:08] <@seb128> you need a GPL copy there in the tarball
  49 [13:08] <@seb128> otherwise the source can't be distributed
  50 [13:08] <@cjwatson> [ldtp] reference: http://people.canonical.com/~cjwatson/ubuntu-policy/policy.html/ch-controlfields.html#s-f-Standards-Version
  51 [13:09] <ara> [ldtp]  cjwatson, oh! thanks! I didn't know that :-) just trying to avoid lintian warnings :-)
  52 [13:09] <@cjwatson> [ldtp] it'd be a good thing to forward to the Debian maintainer, although it isn't really worth a bug report just for that
  53 [13:09] <juanje> [nautilus-md5sum] seb128: Ok. The thing is the software have no upstream tarball or so, the package itself is the upstream, that is whay there is not GPL copy, but I can copy there...
  54 [13:09] <@cjwatson> [ldtp] just as a reminder, one must always check the upgrading-checklist in policy before changing Standards-Version in Debian
  55 [13:10] <@dholbach> [nautilus-md5sum] The build succeeded and I just ran lintian -Iiv nautilus-md5sum_0.3.1-1_amd64.changes which indicated no major mistakes.
  56 [13:10] <@seb128> [nautilus-md5sum] juanje: that doesn't change the fact that it needs the license text in the tarball, yes please do
  57 [13:10] <@cjwatson> [ldtp] Any particular reason you removed yourself from Uploaders?
  58 [13:10] <juanje> [nautilus-md5sum] dholbach: sorry, I think I didn't get the earlier question
  59 [13:10] <@dholbach> [nautilus-md5sum] juanje: are you sure about the python-dev Dependency?
  60 [13:11] <ara> [ldtp] cjwatson, I am uploader in the debian package, but not the ubuntu. so I thought it had to be removed
  61 [13:11] <juanje> [nautilus-md5sum] seb128: Ok, I'll do
  62 [13:11] <juanje> [nautilus-md5sum] dholbach: I think this come up with the pbuilder or something
  63 [13:11] <@seb128> [nautilus-md5sum] juanje: the dirs file is probably not required
  64 [13:12] <@dholbach> [nautilus-md5sum] juanje: the version number in debian/changelog is 0.3.1-1 which indicates that 0.3.1 is the upstream version and -1 the first debian revision. As I see it the package has never been in Debian and this would be its first revision in Ubuntu, so I'd use 0.3.1-0ubuntu1.
  65 [13:12] <@cjwatson> [ldtp] Uploaders means nothing to the Ubuntu archive, so you should leave it in to minimise the delta
  66 [13:12] <juanje> [nautilus-md5sum] dholbach: Ok, that is because I was trying at the same time to get into Debian and Ubuntu
  67 [13:12] <@dholbach> ok
  68 [13:12] <@cjwatson> [ldtp] I'm just checking to ensure that python-wnck is sufficient
  69 [13:13] <@dholbach> [nautilus-md5sum] minor: As the package contains no compiled code, but just a python script, you probably don't need ${shlibs:Depends}.
  70 [13:13] <juanje> [nautilus-md5sum] dholbach: Ok
  71 [13:13] <@seb128> [nautilus-md5sum] juanje: you might want to use pkg-config to get the nautilus directory rather than coding the abi version there
  72 [13:14] <@cjwatson> ./python/ldtpeditor:69:    import gtksourceview
  73 [13:14] <@cjwatson> [ldtp] doesn't that mean you need a dependency on python-gtksourceview as well?
  74 [13:14] <juanje> [nautilus-md5sum] seb128: aha... I'll check
  75 [13:14] <@james_w> anyone else with a package to review?
  76 [13:15] <AndrewGee> I have https://bugs.launchpad.net/ubuntu/jaunty/+source/osm-gps-map/+bug/387043 as a SRU for Jaunty.
  77 [13:15] <@cjwatson> [ldtp] ara: that's everything I can see; if you post a new diff with those changes, I'll sponsor it after this session
  78 [13:15] <ara> [ldtp] cjwatson, thanks for your help!
  79 [13:15] <@dholbach> [nautilus-md5sum] more of a "upstream idea": try using python's distutils and python-distutils-extra for translations, but that's just a suggestion. :-)
  80 [13:16] <juanje> [nautilus-md5sum] I have question about non-native packages, and how to use get-orig/watch file to get a "upstream" tarball and build the package
  81 [13:16] <juanje> [nautilus-md5sum] dholbach: yeah, that's my idea ;-)
  82 [13:16] <@dholbach> [nautilus-md5sum] debian/copyright does not contain a note about where the source code was retrieved from
  83 [13:16] <@dholbach> juanje: just ask :)
  84 [13:16] <@cjwatson> [nautilus-md5sum] I ran across a bug yesterday which had the idea of implementing a get-orig-source sequence in dh(1) with the standard uscan thing ...
  85 [13:17] <juanje> [nautilus-md5sum] dholbach: the code was made for a guy from my team and I like he change into python-distutils-extra the code
  86 [13:17] <@james_w> AndrewGee: that's tricky as it requires an ok from motu-sru before we can upload, and unfortunately for you I'm not in that team
  87 [13:17] <AndrewGee> james_w: Ah. Nevermind then :)
  88 [13:17] <@dholbach> [nautilus-md5sum] juanje: if you have questions about that or need an example, I'm happy to answer that later on in some other place. :-)
  89 [13:18] <arand> Is approvals of queue items in jaunty-proposed in the scope of this session?
  90 [13:19] <@james_w> arand: nope, sorry, that's a task restricted to a particular team
  91 [13:19] <@cjwatson> [osm-gps-map] I'm not convinced about the correctness of adding dpatch in an SRU; I would stick like glue to the current packaging scheme if I were you
  92 [13:19] <@dholbach> [nautilus-md5sum] seb128: Apart from the things we mentioned are you happy with it? :)
  93 [13:19] <juanje> [nautilus-md5sum] The question: I have created debian/watch file and also added get-orig target into the debian/rules file but when I try to build the package the dpkg-buildpackage (or debuild, or pbuilder) cry about not to exist a *.orig directory or tarball, instead of donwload it. Do I need do something before to build?
  94 [13:19] <arand> james_w: okay.
  95 [13:19] <Laney> how about digging into ftbfs? :)
  96 [13:20] <@seb128> [nautilus-md5sum] dholbach: I am yes
  97 [13:20] <@cjwatson> [nautilus-md5sum] get-orig-source is not called manually; it's purely informational. You have to construct the tarball yourself, perhaps by calling 'debian/rules get-orig-source' manually
  98 [13:20] <AndrewGee> [osm-gps-map] cjwatson: How would I go about that? I thought I'd need a patch system.
  99 [13:20] <@cjwatson> [nautilus-md5sum] sorry, I mean "not called automatically"
 100 [13:20] <@cjwatson> [osm-gps-map] you can just edit the source directly
 101 [13:20] <juanje> [nautilus-md5sum] cjwatson: ummmm... I was guessed something like that...
 102 [13:21] <AndrewGee> [osm-gps-map] cjwatson: I thought there wasn't allowed to be any modifications outside the debian/ directory in the deb .diff.gz file? Is this not true?
 103 [13:21] <@dholbach> [nautilus-md5sum] I don't know where you store the code right now, but if you are going to put it into Launchpad and maintain releases there, you could check out   apt-get source --diff-only storm  for an example of a debian/watch file, so you can just use uscan and uupdate afterwards.
 104 [13:21] <@cjwatson> [osm-gps-map] that's a myth
 105 [13:22] <AndrewGee> [osm-gps-map] cjwatson: Okay. I'll get that fixed later :)
 106 [13:22] <AndrewGee> cjwatson: Thanks for your help :)
 107 [13:22] <@cjwatson> [osm-gps-map] furthermore, your patches are the wrong way round at the moment anyway, since aclocal.m4 is (presumably) the source of the configure change and ought to be patched first; but for this I would definitely go with the smallest possible change to the package
 108 [13:23] <juanje> [nautilus-md5sum] dholbach: the code is in LP already. Actually, It was always there: lp:nautilus-md5sum
 109 [13:23] <AndrewGee> [osm-gps-map] cjwatson: Okay. Should I just patch the configure script then?
 110 [13:24] <juanje> [nautilus-md5sum] dholbach: I'll check that package, thanks ;-)
 111 [13:24] <@cjwatson> [osm-gps-map] just to clarify, many people do find patch systems immensely useful; I'm not a fan myself but it's certainly popular among many developers, and if you find a patch system already there you should use it; you should also make your own mind up for new packages. This advice is just for SRUs
 112 [13:24] <@dholbach> [nautilus-md5sum] juanje: ok, I just thought you wanted to put releases there too (0.3.1 somehow indicated there were tarball releases... :-))
 113 [13:24] <@cjwatson> [osm-gps-map] you do need to patch whatever the source of that configure chunk is, not just configure
 114 [13:24] <@cjwatson> [osm-gps-map] otherwise you leave the package in a very confusing state for the next person to come along
 115 [13:24] <@dholbach> what-patch (in ubuntu-dev-tools) can usually tell you quickly which patch system is used by a package. :-)
 116 [13:25] <@dholbach> Do we have some more review requests?
 117 [13:25] <AndrewGee> [osm-gps-map] Okay. Sounds good. So I'll patch aclocal.m4 and configure without dpatch then?
 118 [13:25] <juanje> [nautilus-md5sum] dholbach: Actually we are developing the packages in LP and the upstream code is in hte same branch th
 119 [13:25] <juanje> [nautilus-md5sum] that the debian directory
 120 [13:26] <@cjwatson> [osm-gps-map] for the SRU, right. I'd also recommend at some point switching the order of the patches for Karmic
 121 [13:26] <dutchie> is this the right place to get http://revu.ubuntuwire.com/p/suvat reviewed?
 122 [13:26] <@dholbach> [nautilus-md5sum] juanje: that's fine, then you probably don't need a watch file, just move to 0.3.1 (native package) instead
 123 [13:26] <juanje> [nautilus-md5sum] But I think this is a problem for generating non-native packages automaticaly
 124 [13:26] <AndrewGee> [osm-gps-map] cjwatson: Okay. I'll get that done. Thanks.
 125 [13:27] <juanje> [nautilus-md5sum] dholbach: they were that way before
 126 [13:27] <juanje> [nautilus-md5sum] dholbach: but debian guys ask us to change to non-natives...
 127 [13:27] <@cjwatson> [nautilus-md5sum] I think that's correct, native packages are intended for things that really don't have a use outside of Debian and its derivatives, like dpkg
 128 [13:27] <@dholbach> [nautilus-md5sum] as an upstream it's probably a good idea to roll release tarballs, so other distros can grab them too
 129 [13:28] <@cjwatson> [nautilus-md5sum] they're not intended for cases where it's just a hassle to roll a tarball :)
 130 [13:28] <@dholbach> dutchie: I'll take a look now
 131 [13:28] <@james_w> (tell that to joey :-)
 132 [13:29] <@cjwatson> I thought he just abandoned anything that needed to be non-native :)
 133 [13:29] <@dholbach> [suvat] looks like ./tags is a leftover in suvat_0.2-0ubuntu1.diff.gz
 134 [13:29] <dutchie> [suvat] should it not be included?
 135 [13:30] <@cjwatson> [suvat] it would be unusual to include it just in the diff, yes; the diff should just be what's needed to do the packaging
 136 [13:30] <@cjwatson> [suvat] usually just indicates that somebody forgot to delete it :)
 137 [13:30] <@dholbach> [suvat] I'm not sure what it's purpose is - I just saw a diff to the tags file in suvat_0.2-0ubuntu1.diff.gz
 138 [13:30] <dutchie> [suvat] ok, i know where that's come from
 139 [13:31] <@cjwatson> [suvat] dh_clean removes files called "TAGS", but "tags" is too generic for that
 140 [13:31] <@cjwatson> [suvat] dholbach: it's from tools like exuberant-ctags, for easily following around function names and such in source code
 141 [13:31] <@dholbach> alright
 142 [13:31] <@cjwatson> integrates smoothly with vim
 143 [13:31] <@dholbach> [suvat] 3.8.2 is the current standards-version, you might want to have a look what changed after 3.8.0 in debian-policy
 144 [13:32] <@cjwatson> apt-cache show exuberant-ctags | grep ^Maintainer ;-)
 145 [13:33] <@dholbach> [suvat] you might want to take a look at http://www.debian.org/doc/packaging-manuals/python-policy/ and call dh_pycentral in debian/rules - it'd avoid installing .pyc files in the resulting package
 146 [13:33] <pleia2> hi everyone, james_w just helped me get slightly up to speed so I could get a php package looked at, I'm actually working with the maintainer to review a whole pile of php packages, but I'm not the most experienced reviewer so some help on one of them would be great (they are all pretty similar)
 147 [13:33] <dutchie> [suvat] it's cdbs - debhelper confused me too much
 148 [13:33] <dutchie> [suvat] (my first package)
 149 [13:33] <pleia2> I come from debian packaging, so I haven't gotten into REVU yet (forgive me!) but here's one: http://princessleia.com/deb/html_qf_advmultiselect/
 150 [13:34] <@cjwatson> [suvat] you might consider dh(1) for this package, since it's quite simple; it supports setup.py. You might only need one override target for that symlink at most.
 151 [13:34] <@dholbach> [suvat] you can safely remove debian/{pre,post}{inst,rm} - they don't do anything right now
 152 [13:35] <@cjwatson> [suvat] and if you use dh(1) it will actually get the setup.py install invocation correct to support python2.6
 153 [13:36] <@dholbach> [suvat] to expand on the topic a bit, you just need debian/{pre,post}{inst,rm} if you intend to override the default behaviour during the installation or add something to the process, they will be generated automatically from all the debhelper commands if you choose not change from the default
 154 [13:36] <@cjwatson> [suvat] something like http://paste.ubuntu.com/252473/ I think (untested)
 155 [13:37] <@cjwatson> [suvat] (BTW this is just a suggestion, totally not a review requirement)
 156 [13:37] <@dholbach> [suvat] (minor) lintian suggests to use © instead of (C) in debian/copyright
 157 [13:38] <@cjwatson> james_w: you want to do pleia2's? :)
 158 [13:38] <dutchie> [suvat] I tried something like that. It seems that setup.py installs packages to site-packages when used with --prefix
 159 [13:39] <@dholbach> [suvat] apart from that I think I'm happy
 160 [13:39] <@cjwatson> [suvat] dh_auto_install uses the equivalent of: $(PYTHON) setup.py install --root=$(CURDIR)/debian/$(PACKAGE) --no-compile -O0 --install-layout=deb
 161 [13:39] <@james_w> [php-html-quickform-advmultiselect] couldn't it have a shorter name? :-)
 162 [13:40] <dutchie> [suvat] so I should replace my cdbs-y debian/rules file with the one from the paste?
 163 [13:40] <pleia2> [php-html-quickform-advmultiselect] I'm pretty sure this is the correct naming scheme for these, but yes, this ends up long
 164 [13:40] <@dholbach> [php-html-quickform-advmultiselect] interestingly enough we packaged php-html-quickform for the Ubuntu NGO team a few weeks ago :-)
 165 [13:40] <dutchie> [suvat] get the changes to tags out of the diff.gz
 166 [13:40] <@cjwatson> [suvat] the version I reviewed was not cdbs. It's up to you, you can use cdbs if you like :-)
 167 [13:40] <pleia2> [php-html-quickform-advmultiselect] already I know it needs a standards bump, and there might be a licensing issue (license was fine in older versions, but now lintian is complaining about php license?)
 168 [13:40] <@cjwatson> [suvat] the version I reviewed was just a traditional sequence of dh_* commands
 169 [13:41] <@james_w> [php-html-quickform-advmultiselect] is dh-make-php needed at build time?
 170 [13:41] <pleia2> dholbach: [php-html-quickform-advmultiselect] ah! we were also working on a NMU for php-html-quickform, I'll have to look at what you guys did
 171 [13:41] <@dholbach> [php-html-quickform-advmultiselect] pleia2: https://wiki.ubuntu.com/NGO/ApplicationPackaging
 172 [13:42] <@dholbach> [php-html-quickform-advmultiselect] I'm a bit confused by the php-html-quickform-advmultiselectS/usr/share/php/* files in the .diff.gz - I think they shouldn't be there
 173 [13:42] <@cjwatson> [suvat] oh, I just noticed, for some reason you use dh_python rather than dh_pycentral (to match your build-dependency); I would expect that to print a warning at build-time)
 174 [13:42] <dutchie> [suvat] I changed it to dh
 175 [13:43] <dutchie> [suvat] I'll upload what I've got, hang on a moment
 176 [13:43] <@cjwatson> [suvat] remove that README.Debian, it's not mandatory to have one and it should be user-relevant if you do :-)
 177 [13:45] <@cjwatson> [php-html-quickform-advmultiselect] no need to have stamp files for configure/build targets that don't do anything; the targets could just be empty
 178 [13:45] <@dholbach> [php-html-quickform-advmultiselect] in debian/examples I'd change HTML_QuickForm_advmultiselect-1.5.1/examples/* to HTML_QuickForm_advmultiselect-*/examples/* to not have to update it for every release
 179 [13:45] <pleia2> [php-html-quickform-advmultiselect] specifically the S one? I think you're right, this looks like a mistake
 180 [13:46] <@cjwatson> [php-html-quickform-advmultiselect] that's the sort of thing that happens if you make a typo in the package name in debian/control, and then correct it without first running debuild clean
 181 [13:46] <@cjwatson> BTDTGTTS
 182 [13:46] <@dholbach> [php-html-quickform-advmultiselect] you can remove the usr/share/php/.registry/.channel.doc.php.net/ directory from the package
 183 [13:47] <pleia2> [php-html-quickform-advmultiselect] sneaky dot files, good catch
 184 [13:48] <@cjwatson> any more for any more? I think we're coming close to the end of the currently open reviews
 185 [13:48] <@dholbach> [php-html-quickform-advmultiselect] the debian/watch file does not work - for php-html-quickform I used "http://pear.php.net/package/HTML_QuickForm http://download.pear.php.net/package/HTML_QuickForm-([\d.RC]+).tgz"
 186 [13:48] <dutchie> [suvat] I've uploaded it, might take a few minutes to show up
 187 [13:49] <@dholbach> [php-html-quickform-advmultiselect] pleia2: who are you working with in Debian to get it included? :-)
 188 [13:49]  * ideamonk is away: I'm busy
 189 [13:50] <pleia2> [php-html-quickform-advmultiselect] dholbach: the debian php team in alioth, and seanius but he had the nerve to go on vacation ;)
 190 [13:50] <@dholbach> ah ok :)
 191 [13:51] <@dholbach> ok, do we have any more review requests?
 192 [13:54] <dutchie> [suvat] OK, the upload is showing up now.
 193 [13:54] <@cjwatson> [suvat] did that ln -s in build: that was there before not turn out to be required, then?
 194 [13:55] <dutchie> [suvat] not really
 195 [13:57] <dutchie> [suvat] setup.py puts a .so in build/lib.$ARCH/ that needs to be in . to run the program, that was all
 196 [13:57] <@dholbach> We have ~3 minutes left in the session - are there any comments about the way how we ran the session today? Did you find it confusing, helpful, fun, great, 'orrible? :)
 197 [13:58] <@cjwatson> [suvat] ok, if you feel you need to add it back later then you can use an override_dh_auto_build: target; you need to build-depend on debhelper (>= 7.0.50) if you do that
 198 [13:58] <dutchie> [suvat] is it otherwise ok as it is?
 199 [13:59] <alourie|work> dholbach: great :-)
 200 [13:59] <@dholbach> so something we should repeat? :-)
 201 [14:00] <dutchie> dholbach: definitely
 202 [14:00] <@dholbach> thanks a lot to the reviewers and those who plucked up the courage to put up their review request!
 203 [14:00] <@dholbach> have a great day!
 204 [14:00] <pleia2> thanks!
 205 [14:00] <@cjwatson> thanks, everyone
 206 [14:01]  * dholbach will put up the logs later on
 207 [14:02] <@seb128> thanks
 208 


CategoryPackaging

Packaging/Training/Logs/2009-08-13 (last edited 2009-08-13 14:16:37 by dholbach)