On 12-Sep-00 at 10:12, Alexander Mai ([EMAIL PROTECTED]) wrote:
> Forgot to mention some things
>
> On Tue, Sep 12, 2000 at 05:55:13AM -0400, Rick Scott wrote:
> > On 12-Sep-00 at 05:19, Alexander Mai ([EMAIL PROTECTED])
> > wrote: > On Mon, Sep 11, 2000 at 09:46:54PM -0400, Rick Scott wrote:
> > > > I finally got around to getting openMotif to compile on my box. Okay,
> > > > I had to #if 0 out a wack of stuff dealing with XICProc's, whatever
> > > > they are, so these results may be tainted a bit. Anyway, here is what
> > > > I found......
> > > > With LessTif
> > > > we fail 63 out of 555 tests
> > > >
> > > > With openMotif
> > > > we fail 92 out of 519 tests
> > > >
> > > > So to start 36 of the tests did not even compile under openMotif.
> > > > Correct me if I'm wrong Jon, all of the tests have compiled, and
> > > > run, under some version of Motif 1.2.
> > > >
> > > > Unfortunatly, all is not rosy.....We are segfaulting on _alot_ of
> > > > tests that we never used to!!!!!! For example, textf/test8!!!!!
> > > > Simply creating and destroying a damn TextField, we are now
> > > > segfaulting!!!!
>
> Fixed.
> Seems it was the patch which I applied earlier (well, not my patch, but my
> fault not to run checks on it probably)
> Now trying at least to locate the filesb problem.
>
> > Also
> > > > filesb test11, 12, an 13!!!! This is just damn depressing...... A
> > > > hell of a lot of time and effort have gone into the tests over
> > > > _many_ years, and for what?? It seems that they are not being used
> > > > anymore.....
> > > > So much for improvements.....the next step needs to be going back and
> > > > get running that which ran before......sigh.....
>
> The tests are helpful, but if not running from a test facility like you did
> earlier things are more difficult perhaps.
>
> Ok, what about me?
> - my OS/2 port is still based on some manual written Makefiles
> test/ is not supported yet ...
Then one of our priorities should be to get this working. Nothing is stopping
us from putting
if [ $USER = amai -a `uname -n` = "myos2box" ]
then
fi
into the test script, or even into the tests themselves. The main benefit of
the tests is to know what passes and what fails before a patch, and after the
patch. If something fails after a patch that used to pass, it is not
necessarily wrong, but needs further investigation. The recent BaseClass
regression is a good example. It initially caused many of the previously
passing tests to fail, but after investigation it was the right thing to do,
and fixes were applied in other areas.
> - my linux box is too weak to make running testall real fun :-(
This is where Xfvb comes in handy. Do a ./testall *, then icon the xterm that
you started it in and take a look in the morning.
> - my DU box: well, here I build the tests and run some. Not regularly
> though. WRT Motif tests: I also only built single ones, I even don't
> know if there's a generic Makefile target for building all Motif tests?
Well, in each of the bottom level directories, "make motif-tests" will build
all of the test*.motif files. I'm starting to think that a target one level up,
in the test/Xm directory would probably be a good thing..... Of course you end
up needing lots of space for this.....
>
> --
> Alexander Mai
> [EMAIL PROTECTED]
>
>
