On Sat, Jul 04, 2015 at 05:21:01PM +0200, Alexandre Detiste wrote: > > thanks for caring about reproducible builds and filing bugs against your > > own > > packages! :-) > > I got conviced (and a bit entertained) by the talk at FOSDEM ;-)
;D > I think that Mattia's guess was closer: eheh, that was really a guess! /me goes making some bets.... > >> Quite some FTBFS on our infrastructure are do to us using > >> DEB_BUILD_OPTIONS=parallel=23 (e.g. very high) and some packages not coping > >> fine with it. > > I finally found ou the likely cause: > 'install' target in Makefile.in is split in tiny chunks. I didn't look at any way at the package other than the build log. > I don't know if it still make any sense, this was already done that way > before the SVN -> Git transition from 2005, and I'm maintaining this > for less than a year. > > "install : install1 install11 install2 installdirs cruft" > > If 'install11' run before 'intall1'; it fails. > Adding a simple sleep 0.1at the top of "intall1" always trigger > an error in "install11". > > Adding an extra "mkdir -p $(DESTDIR)/usr/lib/cruft/" fix this permanently (?) > > https://github.com/a-detiste/cruft/commit/4e8a48999bd1c77c0d902c803f4151f3c09f471a well, maybe if they are too tiny even mergning them can make sense > > > To achieve reproducibility you will also need to normalize the timezone > > during > > build (eg set TZ=UTC) or wait til debhelper does this for you. > > All the previous run that didn't FTBFS were reproducible; > are there any other changes needed ? umh, indeed. Holger? -- regards, Mattia Rizzolo GPG Key: 66AE 2B4A FCCF 3F52 DA18 4D18 4B04 3FCD B944 4540 .''`. more about me: http://mapreri.org : :' : Launchpad user: https://launchpad.net/~mapreri `. `'` Debian QA page: https://qa.debian.org/developer.php?login=mattia `-
signature.asc
Description: Digital signature
