Re: [Libreoffice-commits] core.git: Let's have activex control test also a slowcheck test

Hi Stephan,

I was how added this activex test to the source. First I added it to
subsequent_test because I don't want it to run by make <module>, but then I
realized that there is also the slowchecks, which are not run by make
<module>, but run by top level make. It has the same dependency as other
filters test in sd.

Best Regards,
Tamás

2017-11-13 15:43 GMT+01:00 Stephan Bergmann <sbergman@redhat.com>:

> On 11/13/2017 11:54 AM, Tamás Zolnai wrote:
>
> > commit 0c5eda9876f353e0516171fec67568643f24c5b6
> > Author: Tamás Zolnai <tamas.zolnai@collabora.com>
> > Date:   Mon Nov 13 10:13:13 2017 +0100
> >
> >      Let's have activex control test also a slowcheck test
> >           Change-Id: I55f8f0ca9478e676832ebbb08e98dbd1cf0fa4fc
> >      Reviewed-on: https://gerrit.libreoffice.org/44666
> >      Tested-by: Jenkins <ci@libreoffice.org>
> >      Reviewed-by: Tamás Zolnai <tamas.zolnai@collabora.com>
> >
> > diff --git a/sd/Module_sd.mk b/sd/Module_sd.mk
> > index becd4528e81f..59f792a78037 100644
> > --- a/sd/Module_sd.mk
> > +++ b/sd/Module_sd.mk
> > @@ -37,6 +37,7 @@ $(eval $(call gb_Module_add_slowcheck_targets,sd,\
> >       CppunitTest_sd_filters_test \
> >       CppunitTest_sd_misc_tests \
> >       CppunitTest_sd_html_export_tests \
> > +    CppunitTest_sd_activex_controls_tests \
> >   ))
> >   endif
> >   @@ -56,7 +57,6 @@ $(eval $(call gb_Module_add_screenshot_targets,sd, \
> >     $(eval $(call gb_Module_add_subsequentcheck_targets,sd,\
> >       JunitTest_sd_unoapi \
> > -    CppunitTest_sd_activex_controls_tests \
> >   ))
> >     # vim: set noet sw=4 ts=4:
>
> "Subsequent checks" were originally introduced to work around dependency
> issues:  Those tests depended on the instdir (or its equivalent, back then)
> installation set being fully populated, so that those tests (mostly the
> infamous qadevOOo-style JunitTests, which spawn a full soffice process)
> could expect everything to be available without specifying full
> dependencies.  (Where specifying the full dependencies would not even have
> been possible in all cases, IIRC.)
>
> I am not sure how much of that is still relevant, what (if any) implicit
> dependencies a subsequentcheck can still implicitly rely on.  However, be
> careful when moving a test from subsequentcheck to somewhere else. It might
> still carry such implicit dependencies.
> _______________________________________________
> LibreOffice mailing list
> LibreOffice@lists.freedesktop.org
> https://lists.freedesktop.org/mailman/listinfo/libreoffice