Today I didn't review all yours remarks. But in the spirit of release early and release often here goes my today effort. On 07/02/16 22:01, Mattia Rizzolo wrote: > On Sun, Feb 07, 2016 at 08:40:10PM +0000, Jose M Calhariz wrote: >> On 06/02/16 23:41, Mattia Rizzolo wrote: >>> Umh, couldn't you turn d/rules to use the dh sequencer? >> I don't know enough and lintian show many problems with upstream d/rules. > well, let's fix them, then :) > Attached there is a d/rules using short dh, may you try it and bend it > better to the needs of this package? I managed to make it compile, but for a reason I don't know I needed to add the following lines: override_dh_auto_configure: cp /usr/share/misc/config.guess . cp /usr/share/misc/config.sub . dh_auto_configure --parallel I have tried many ideas but was only this way that it build > There are some errors, for example lintian complains about > dependency_libs in the .la files [0], and other stuff. > though it also fixes things like debian-changelog-file-missing and > no-copyright-file I otherwise have with the current rules file. > Another thing: the build log looks really different, there are way more > lisp-related lines, but I haven't looked at it at all. > And it's really untested and I think it would enjoy some more work. > >>> I anyway see a lot of nowadyas obsolete stuff that could really easily >>> be removed, for example: >>> >>> * all that if/else/endif staff can go away if you stop calling manually >>> ./configure and rely in dh_auto_configure >> Done > though you didn't remove the if/else/endif at the start of d/rules. Gone > >>> * the source targe should follow policy §4.9 (so, named get-orig-source) >> Removed, not usefull for me. > cool. > >>> * d/changelog >>> + "Remove upstream debian directory." => what's this? do you mean the >>> upstream tarball and you removed it? source fomat 3.0 already does >>> it by itself, i don't understand what you did. >> I created a commit in git that removes the upstream debian directory. >> Maybe is not necessary. > umh now I don't recall if gbp does the wrong thing while importing the > ustream tarball and do things wrong, but I hope it doesn't. Anyway, > that's a kind of thing that is not thought to be in d/changelog. > >>> * d/control >>> * maybe you can consider to drop the -dbg package in favour of the >>> automatically built dbgsym? >> Done > cool! > >>> * d/menu: drop it? >> Why? > https://lintian.debian.org/tags/command-in-menu-file-and-desktop-file.html Done > >>> * d/copyright: consider write a copyright-format 1.0 one? at a first >>> sight doesn't look too much work. >> Done > though it's not compliant, and indeed lintian is noisy on it, please try > to figure out what's wrong with it and fix it. > I believe blindly following lintian here is enough, though it would be > nice if you could understand what's the problem by yourself :) I fix it, but I don't understand why :-) > >>> * once fixed the b-d, I get another ftbfs, with a lisp backtrace ending: >>> 'error--> (file-error "No such file or directory" "rep/data/tables")' >>> go figure... >> Are you building in i386 or amd64? > amd64. > though now it builds, umh. ok... I found a missing bump on Build-depens on rep. > >>> * so it seems you have symlinked doc dir. do as it pleases you, but >>> anyway please drop those manually built .postinst file if their only >>> use it that one. >> Droped > I haven't yet checked, though you need to make sure there is an upgrade > path, and remember that dpkg doesn't overwrite a symlink while > installing/upgrading a package. do you remember those symlink_to_dir > and dir_to_symlink thinghies of the other package? I remember something. I have to review it. > >>> * please try to get a reproducible buildable package, from what I see it >>> wouldn't be difficult at all. >>> >>> there are 57 open bugs, are you telling me none of them get closed by >>> this upload? :\ >>> I anyway expect some bug triaging done… >> Will do it. > > * d/clean: you are removing config.sub, config.guess, configure and > libtool. Well, > 1) theoretically once the package is built the clean target of d/rules > should bring the tree back to the original state. > 2) it makes git unnecessarly noisy about deleted files from the tree. > Just removing those 4 files from the d/clean file is enough, and > anyway I don't understand what would have been the whole point of > having them listed there in the first place. > > * P: sawfish-lisp-source: maintainer-script-without-set-e postinst > + and more like that > * d/rules: please try enabling the hardening build flags > > > > [0] add here a random rant about static libaries...... can't we just > drop them... :\ (last time I brought this up others said to keep > them for our users...... gah!)
Attachment:
signature.asc
Description: OpenPGP digital signature