Re: librtr: Request for review / upload (Re: Git repository for librtr)

[Lukas Schwaighofer <lukas@schwaighofer.name>]

Hi Samuel,

thanks for looking at this again and sorry for the delay in my response
(I've been traveling…).

On Sat, 3 Nov 2018 04:50:00 -0300
Samuel Henrique <samueloph@debian.org> wrote:

> > d/watch: ok  
> > > d/control: ok, but it's better with wrap-and-sort  
> >
> > I don't really like the warp-and-sort style for listing the
> > Build-Dependencies.  Having the first b-d on a new line and all of
> > them with a trailing comma gives a lot more readable git diffs. If
> > you insist I'll change it through.
> >  
> 
> Not at all, I won't complain if you don't want to, personally I
> rather have it one per line because I find the git diff better this
> way, I thought you'd find it too.

To clarify what I meant, I prefer

Build-Depends:
 dep1,
 dep2,
 dep3,

to what wrap-and-sort does, which is

Build-Depends: dep1,
               dep2,
               dep3

because it will also give me a clean diff when I do something to the
first or last dependency.

I don't care too much about all the other Depends: fields since I don't
expect them to ever change. misc:Depends and shlib:Depends are
awesome :).

> The package is FTBFSing for me randomly, I'm building it with sbuild
> on a amd64 with 8 cores and 16gb ram, here is the error:
> [log removed]
> I tried building a few times to get the probability of a failed build
> and got 6 fails out of 13 attempts.
> 
> I think this will have to be fixed before uploading the package, let
> me know if you're able to replicate on your machine.

I've a quite similar build system and I haven't been able to reproduce
the problem.  I've probably done around 50 builds by now (I use
`sbuild`).  I've also just now run only the test suite 100 times
without encountering any failed tests.

Looking at the verbose log you provided, it looks exactly like a
successful run except that it stops at some point.  Does the log always
look the same?


In any case, I've also talked to upstream and the best guess is that
this is some kind of network issue.  Both the live_validation and
dynamic_groups test make a connection to upstream's server, which means
they would fail on the buildd's as well.  I've disabled those two tests
for that reason and pushed the change to git.

Thanks
Lukas