[QGIS-Developer] Is qgis_sip_uptodate test giving false negative ?

[Denis Rouzaud]

I think it's the sip_include test which is causing issues.

Le mar. 1 oct. 2019 à 12:38, Sandro Santilli <strk at kbt.io> a écrit :

> Ok sorry for the noise, the OptName=s syntax is to specify that
> the option takes an argument, doesn't give an indication of a
> corresponding short-version option. I'm still trying to understand
> why the server_auto.sip file is modified by travis...
>
> --strk;
>
> On Tue, Oct 01, 2019 at 12:31:13PM +0200, Sandro Santilli wrote:
> > Denis, you seem to have introduced both the new switch (with the
> > typo) and the corresponding call with commit
> > 6301f9b6f3ba1c9b39ffe55698365ab98aa059a1 on May 2018.
> >
> > I'm not sure how this could have gone unnoticed till now,
> > maybe the sipify.pl sript should noisily fail when passed
> > unknown switches ?
> >
> > --strk;
> >
> > On Tue, Oct 01, 2019 at 12:28:39PM +0200, Sandro Santilli wrote:
> > > Looking at Travis output, the qgis_sip_uptodate test passes:
> > >
> > >   https://travis-ci.org/qgis/QGIS/jobs/591915632#L958
> > >
> > > BUT a test I've added to check if repository files changed
> > > shows that the `python/server/server_auto.sip` file in the
> > > git repository was changed during tests:
> > >
> > >   https://travis-ci.org/qgis/QGIS/jobs/591915632#L941-L944
> > >
> > > Are the two facts contraddictory ?
> > >
> > > Reading scripts/sipify.pl it looks like the `-p` switch is
> > > not supported, while the `-s` switch is used for two different
> > > switches (a typo?):
> > >
> > >     GetOptions ("debug" => \$debug, "sip_output=s" => \$sip_output,
> > >
>  "python_output=s" => \$python_output) && @ARGV == 1;
> > >
> > > I mention this because tests/code_layout/test_sipfiles_uptodate.sh
> > > uses the -p switch instead:
> > >
> > >             ./scripts/sipify.pl -p
> python/${module}/auto_additions/${pyfile}.temp $header |
> > >                             diff python/$sipfile.in -
> > >
> > > So, I suspect sipify.pl is _overriding_ the source file and NEVER
> producing
> > > a .temp file, resulting in:
> > >
> > >   1) A false OK message from Travis (no difference!)
> > >     2) A modification of source tree during test
> > >
> > > --strk;
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.osgeo.org/pipermail/qgis-developer/attachments/20191001/64390715/attachment.html>