Re: Running tests under wine.

[Ben Pfaff]

OK, I'll work on that.

On Tue, Mar 22, 2016 at 09:26:44PM +0100, Harry Thijssen wrote:
> It would be worth a try.
> 
> Could you provide me with a script for these tests and instruct me how to
> use it?
> 
> Have fun
> 
> 2016-03-22 21:22 GMT+01:00 Ben Pfaff <address@hidden>:
> 
> > On Tue, Mar 22, 2016 at 09:04:07PM +0100, Harry Thijssen wrote:
> > > 2016-03-22 16:23 GMT+01:00 Ben Pfaff <address@hidden>:
> > >
> > > > On Tue, Mar 22, 2016 at 06:56:17AM +0100, John Darrington wrote:
> > > > > On Mon, Mar 21, 2016 at 10:55:19PM +0100, Harry Thijssen wrote:
> > > > >      > Date: Sun, 20 Mar 2016 21:59:02 +0100
> > > > >      > From: John Darrington <address@hidden>
> > > > >      > To: address@hidden
> > > > >      > Subject: Running tests under wine.
> > > > >      >
> > > > >      > I pushed a change which allows many of the tests to run under
> > > > wine.
> > > > >      >
> > > > >      > make RUNNER=wine check
> > > > >      >
> > > > >      > --
> > > > >      >
> > > > >
> > > > >      I tried it  and see the results below. Due to a patch of mine,
> > > > there should
> > > > >      1 tests and 1 failure added to the counts.
> > > > >      Would the same trick work for some of the other executables
> > used in
> > > > the
> > > > >      tests?
> > > > >
> > > > > Unfortunatey I don't think so.  For two reasons:
> > > > >
> > > > > 1.  We cannot "shadow" the pspp-convert command using a bash
> > function,
> > > > because
> > > > >     bash functions cannot contain hyphens.
> > > >
> > > > pspp-convert is much less of a problem, because I only see a few calls
> > > > to it.  It's easy enough to add $RUNNER directly to those.
> > > >
> > > > > 2.  As discussed earlier, many of the other commands need to be
> > compiled
> > > > for the
> > > > >     build machine rather than the host machine.  This will require
> > > > significant
> > > > >     rearrangement of the automake files.
> > > >
> > > > Yes, this is more of a problem.
> > > >
> > > >
> > > I did a quick check and found:
> > > datasheet-test.exe
> > > inexactify.exe
> > > sack.exe
> > > command-name-test.exe <http://lexer.at>
> > > scan-test.exe
> > > segment-test.exe
> > > abt-test.exe
> > > bt-test.exe
> > > cmac-aes256-test.exe
> > > encoding-guesser-test.exe
> > > heap-test.exe
> > > hmap-test.exe
> > > hmapx-test.exe
> > > i18n-test.exe
> > > line-reader-test.exe
> > > ll-test.exe
> > > llx-test.exe
> > > range-map-test.exe
> > > range-set-test.exe
> > > range-tower-test.exe
> > > sparse-array-test.exe
> > > sparse-xarray-test.exe
> > > stringi-map-test.exe
> > > stringi-set-test.exe
> > > string-map-test.exe
> > > string-set-test.exe
> > > str-test.exe
> > > tower-test.exe
> > > u8-istream-test.exe
> > > zip-test.exe
> > > chart-get-scale-test.exe
> > > chart-get-ticks-format-test.exe
> > > render-test.exe
> > > syntax-gen-test.exe
> > >
> > > Wouldn' t it work when $RUNNER is used in the calls for these programs?
> > > Maybe I can run a test a see how many of the tests complete then.
> >
> > That would probably help a great deal.
> >
> > It would be pretty easy to generate redirection scripts for these
> > programs, similar to how we generate redirection scripts when we run
> > "check-valgrind".  That might solve most of the problems with
> > cross-checking.
> >