[Spice-devel] [PATCH spice-streaming-agent v3 3/3] mjpeg-fallback: unittest for the options parsing

[Victor Toso]

On Tue, Feb 20, 2018 at 12:22:52PM +0100, Lukáš Hrázký wrote:
> On Tue, 2018-02-20 at 11:59 +0100, Victor Toso wrote:
> > On Tue, Feb 20, 2018 at 11:45:33AM +0100, Lukáš Hrázký wrote:
> > > > That tests are not enabled by default. If we enable it by default
> > > > here I would expect to do the same for other Spice components.
> > > 
> > > Ok, but I'm somewhat confused here. Enabling the tests is during
> > > packaging - you run `make check`. Which means we are talking about...
> > > Fedora packaging? Or am I getting something wrong?
> > 
> > I'm talking about releases and the configure script with default
> > options. Today we don't have tests enabled by default in our
> > tarballs.
> > 
> > 1) wget https://www.spice-space.org/download/gtk/spice-gtk-LATEST.tar.bz2
> > 2) tar -xvf spice-gtk-LATEST.tar.bz2
> > 3) cd spice-gtk-0.34
> > 4) ./configure
> > 5) make
> > # no tests, no extra dependencies.
> 
> Thanks for the example, made it clear. I think you are mistaken here,
> though:
> 
> wget https://www.spice-space.org/download/gtk/spice-gtk-LATEST.tar.bz2
> tar -xvf spice-gtk-LATEST.tar.bz2
> cd spice-gtk-0.34
> ./configure
> make check
> 
> (the only difference being calling `make check` at the end)
> 
> Here you have the tests.

For spice-common (submodule) not spice-gtk. For spice-gtk,
you need --enable-static and then, yes, make check will run tests
for it.

> the tests. Many packaging tools automatically detect there is a make
> check target and call it at the appropriate point during the packaging.
> 
> So you don't need to explicitly enable the tests in the tarball in any
> way. The only issue, which we are discussing here, are extra testing
> dependencies, that the tests may have and that you check for in
> configure.
> 
> So if someone builds the package and does not intend to run the tests,
> the configure script may still require him to install the tests
> dependencies.
> 
> And as I break it down here for myself as well, I think I was wrong
> with my previous argumentation for the different behaviour based on
> presence of Catch in the system. Because the make check target will be
> there anyway, it will just be broken if the dependency isn't there.
> Correct? So in the unlikely case I'm not missing anything anymore, I
> suppose Christophe's suggestion is ok. Sorry for the noise :)

Yes, warning is fine by me too.
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 833 bytes
Desc: not available
URL: <https://lists.freedesktop.org/archives/spice-devel/attachments/20180220/d142e444/attachment.sig>