On Mon, Feb 18, 2019 at 09:25:55AM +0100, Ricardo Wurmus wrote:
> Leo Famulari <leo@famulari.name> writes:
> > * gnu/packages/python-crypto.scm (python-txtorcon): New variable.
> […]
> > +    (arguments
> > +     `(#:tests? #f)) ; fail silently
> 
> Could you please elaborate on this?  Do all tests fail?  Would it make
> sense to disable the broken tests?  Are test failures known to upstream?

Basically, the first test that runs looks for a file
'release-announce-test' in the root of the tarball.

This file does not exist in the PyPi or GitHub releases (they are
identical). I checked several releases going back to December 2015, and
none of them contain this file, so I'm not sure what's going on.

Reported upstream at <https://github.com/meejah/txtorcon/issues/330>.

I'll add a comment explaining this.

Also, while skimming some other tests I began to suspect they require
internet access, although I'm not sure.