Re: [PATCH]: gnu: gnurl: Update to 7.50.1.

[Leo Famulari <leo@famulari.name>]

On Sun, Aug 14, 2016 at 09:18:11AM +0000, ng0 wrote:
> ng0 <ng0@we.make.ritual.n0.is> writes:
> > ng0 <ng0@we.make.ritual.n0.is> writes:
> >> Leo Famulari <leo@famulari.name> writes:
> >>> Can you add a comment explaining why test1139 is disabled? Just saying
> >>> that it won't work doesn't tell those reading the package why it is
> >>> disabled.
> >>
> >> My assumption: files or lines of code are searched which will not exist
> >> in gnurl. If this is not just us (sadly in Gentoo i have test/checks
> >> disabled for this package), I will fix this upstream.
> >> If your perl knowledge is better look at the failing test yourself. This
> >> is my best judgment with the little perl knowledge I have.
> >> It was just ruling out what could cause the test failure.
> >
> > Is this okay?
> 
> Can someone tell me if this is okay or if it needs further changes?
> I'd like to update this.

Okay, I think it should updated too, considering that gnurl is a cURL
fork, and that the latest release of cURL fixed these bugs:

CVE-2016-5419:
https://curl.haxx.se/docs/adv_20160803A.html
CVE-2016-5420:
https://curl.haxx.se/docs/adv_20160803B.html
CVE-2016-5421:
https://curl.haxx.se/docs/adv_20160803C.html

Pushed as 5f9d5905745.

Will you file a bug report with the gnurl developers about this failing
test and reply to this email with a link to the bug report?

Ideally, the package update commit would include a link to the upstream
bug report about the failing test, but I'm not going to wait for the
link to update this.

If a Guix package is passing its entire test suite, we should react to
new tests failures proactively. That means that we should try to
understand the problem, and work with the upstream developers to fix the
problem.