[libvirt PATCH 246/351] meson: src: add check-augeas test

[Peter Krempa]

On Tue, Jul 28, 2020 at 10:09:11 +0200, Pavel Hrdina wrote:
> On Tue, Jul 28, 2020 at 09:51:40AM +0200, Peter Krempa wrote:
> > On Thu, Jul 16, 2020 at 11:58:02 +0200, Pavel Hrdina wrote:
> > > Signed-off-by: Pavel Hrdina <phrdina at redhat.com>
> > > ---
> > >  scripts/check-augeas.sh | 12 ++++++++++++
> > >  scripts/meson.build     |  1 +
> > >  src/Makefile.am         | 17 -----------------
> > >  src/meson.build         | 18 +++++++++++++++++-
> > >  4 files changed, 30 insertions(+), 18 deletions(-)
> > >  create mode 100644 scripts/check-augeas.sh
> > 
> > [...]
> > 
> > > @@ -857,3 +860,16 @@ test(
> > >    args: [ check_aclrules_prog.path(), files('remote/remote_protocol.x'), stateful_driver_source_files ],
> > >    env: runutf8,
> > >  )
> > > +
> > > +if augparse_prog.found()
> > 
> > Can't we use foreach on augeas_test_data and invoke the test
> > individually for each file rather than adding a script which does the
> > same?
> 
> Sure we can, I wanted to stay as close as to autotools where
> check-augeas is a single target. Using foreach would result into having
> separate target for each file. In general it is most likely better but
> I was trying to avoid any changes like this with the rewrite and do them
> as a followup patches to cleanup the build system.

In autotools the loop is in the automake file (yes, still shell),
while here you put it into a separate script file. I'd argue that
putting the loop into the meson file (regardless of how many test
targets it creates) is more equivalent to the original source than this
way.

Given that you are now changing all the shell helpers to python I'd
rather see this done in pure meson, than having a python file looper
intermediate helper.