Re: [lmi] CI builds fixed again

[Vadim Zeitlin]

On Mon, 14 Feb 2022 21:55:29 +0000 Greg Chicares <gchicares@sbcglobal.net> 
wrote:

GC> On 2022-02-13 01:55, Vadim Zeitlin wrote:
GC> > 
GC> > [...] simply launching Xvfb in a
GC> > separate build step from running the tests themselves was enough to make
GC> > them work, unlike everything else that I tried.
GC> 
GC> In my experience as well, the interaction between Xvfb and wine is
GC> complicated and finicky. Glad you found a way to make this work.

 I don't think it's just due to Xvfb and wine, as I could never reproduce
the problem locally (and not for the lack of trying). Something must
interfere with it on GitHub VM/container, but I have no idea what as even
my wildest hypotheses (and I honestly am afraid to admit the things I've
tried) couldn't explain it. But I'll just have to live with this mystery.

GC> >  So I've now pushed my changes to master and all the CI builds pass again
GC> > in it now, see the latest build at
GC> > 
GC> >   https://github.com/let-me-illustrate/lmi/runs/5170674974
GC> > 
GC> > [...] now we
GC> > have CI builds using both 32 and 64 bit MinGW versions.
GC> 
GC> Thanks.
GC> 
GC> At the 'github' link above, I see two things that still seem to fail:
GC> 
GC>  - "Configure lmi" for both "MSW/make/mingw32" and "Linux/make/gcc"; and
GC> 
GC>  - "Launch Xvfb" and "Run GUI tests" for "Linux/make/gcc", where I
GC>    suppose Xvfb would work just fine, whereas the GUI tests don't; but
GC>    IIRC that's on your TODO list already.
GC> 
GC> Is there anything I can do to help with either of those?

 These are not failures, but rather these steps are (intentionally)
skipped. The "Configure lmi" one is skipped just because this step is done
only for the builds using autoconf and so it doesn't make sense for the
makefile-based builds, but the steps mentioned under the second bullet
point above are indeed only skipped because running GUI tests would
currently fail. I'd like to this, but unfortunately I haven't even started
looking at this yet (except for quickly coming to the conclusion that this
is not going to be that simple to fix, unfortunately), due to all my time
being spent on high DPI stuff, and I'm afraid that if I do it now, it will
delay wx upgrade even more.

 So, to answer your question, the first thing doesn't need to be fixed and
while the second does, I'm afraid I need to do it at wxGTK level.

 But thanks for asking!
VZ

pgp9jwCExQ8O7.pgp
Description: PGP signature