[Pkg-netatalk-devel] 3.1.16~ds-1

[Daniel Markstedt]

------- Original Message -------
On Saturday, September 16th, 2023 at 12:24 PM, Daniel Markstedt <daniel at mindani.net> wrote:


> 
> 
> 
> 
> 
> 
> 
> ------- Original Message -------
> On Tuesday, September 12th, 2023 at 10:29 AM, Jonas Smedegaard jonas at jones.dk wrote:
> 
> 
> 
> > Quoting Daniel Markstedt (2023-09-12 18:06:37)
> > 
> > > On Tuesday, September 12th, 2023 at 6:23 AM, Jonas Smedegaard jonas at jones.dk wrote:
> > > 
> > > > Quoting Daniel Markstedt (2023-09-12 09:05:39)
> > > > 
> > > > > I have prepared the code changes for 3.1.16~ds-1, with this changelog:
> > > > > https://salsa.debian.org/netatalk-team/netatalk/-/blob/debian/latest/debian/changelog
> > > > > 
> > > > > It should be ready for testing, except I think I messed up the pristine tarball delta generation.
> > > > > The binary delta came out way too large, and debuild complains about binary files that has changed.
> > > > > So I think I made a mistake somewhere working at this too late at night!
> > > > > 
> > > > > In hindsight I should not have pushed to the pristine-tar remote before fully testing:
> > > > > https://salsa.debian.org/netatalk-team/netatalk/-/blob/pristine-tar/netatalk_3.1.16~ds.orig.tar.gz.delta
> > > > > So I guess we may have to revert this and start over. Sorry about the mess!
> > > > > I'll try to look at it again later in the week.
> > > > > 
> > > > > This is what debuild complains about:
> > > > > dpkg-source: error: cannot represent change to doc/gfx/logo.png: binary file contents changed
> > > > > dpkg-source: error: add doc/gfx/logo.png in debian/source/include-binaries if you want to store the modified binary in the debian tarball
> > > > > dpkg-source: error: cannot represent change to etc/afpd/spotlight-packet.bin: binary file contents changed
> > > > > dpkg-source: error: add etc/afpd/spotlight-packet.bin in debian/source/include-binaries if you want to store the modified binary in the debian tarball
> > > > > dpkg-source: error: cannot represent change to etc/afpd/spotlight-packet2.bin: binary file contents changed
> > > > > dpkg-source: error: add etc/afpd/spotlight-packet2.bin in debian/source/include-binaries if you want to store the modified binary in the debian tarball
> > > > > dpkg-source: error: unrepresentable changes to source
> > > > 
> > > > The above indicates either a totally broken set of pristine-tar +
> > > > upstream/latest git branches or an obsolete local auto-generated
> > > > tarball that you simply need to delete to have git-buidpackage generate
> > > > a fresh one properly in sync with corresponding git branches/tags. Try
> > > > the latter first, and if that doesn't work then tell me, and I recommend
> > > > that I try clean up the mess (because it involves either force-pushing
> > > > public git branches or merging to paper over the mess, where I prefer
> > > > the former because the latter can later bite us in the ass due to
> > > > subtleties of how git tries to second-guess evolution of changes.
> > > > 
> > > > - Jonas
> > > 
> > > You're right, I forgot about the `pristine-tar gentar` step.
> > > Unfortunately, my critical mistake here, I think, is that I generated the delta before removing all of the generated in/m4/h files.
> > > They're included in the upstream tarball but have been purged in upstream/latest ...
> > > This must be why the binary delta is much larger than expected.
> > > 
> > > So `pristine-tar gentar` fails with the delta because it cannot stat the various Makefile.in files etc.
> > > 
> > > Please let me know if there's anything else I should try before we (you) try to clean this up...
> > 
> > Hmm, it sounds like you are doing far too much by hand. I've never
> > heard about a "pristine-tar gentar step" but am not surprised that
> > somewhere deep down in the layers of wrappers around dpkg-buildpackage
> > such command gets called: What I do is eesentially "gbp import-orig"
> > which takes care of piecing it all together - and even handles rolling
> > back the git repo to a sane previous state if it detects that something
> > is off, leaving to hand-holding only situations where it did not
> > notice something being off but I change my mind and want a "succesful"
> > import reverted.
> > 
> > I don't have time today (need to read up for a lecture tomorrow) but
> > will try find time tomorrow to look at it.
> > 
> > Please do ping me if you don't hear from me for several days... :-)
> > 
> > - Jonas
> 
> 
> Hi Jonas!
> 
> ...ping ;)
> 
> In fact, change of plans: Moments ago I tagged and published netatalk v3.1.17 which includes a 0-day patch.
> https://github.com/Netatalk/netatalk/releases/tag/netatalk-3-1-17
> 
> What I think is the best course of action is...
> - Discard 3.1.16~ds-1 (or maybe leave it in the changelog as UNRELEASED?)
> - Prepare a 3.1.17~ds-1 package
> - File debian security bugs against oldoldstable, and oldstable with debpatches attached
> 
> ...anything else?
> 
> I'm about to head out the door for a going-away party before I leave this country, so I might not have time today to do any of this.
> And tomorrow we're visiting someone else. The earliest I can take action is Sunday night my time.
> If you have time this weekend don't hesitate to take action! :)
> 
> Cheers!
> 
> Daniel

For starters I filed this so that the Security team gets a heads-up...

https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=1052087