On Tuesday 27 January 2004 23:56, Rafael Laboissiere wrote:
| * Alan W. Irwin <irwin@...> [2004-01-27 13:18]:
| > Can you confirm that hang for plplot-test.sh?
| >
| > It is possible some interactive stuff has been put into x01c.m for RC2
| > which makes it impossible to run it with -dev psc (which is what
| > test_octave.sh attempts to do). If this is the problem, can you change
| > x01c.m to work once again with file devices?
|
| I confirm it here for RC2. Also, when the old RC1 x01c.m is used, the
| hang disappears.
|
| The patch to fix this problem is quite trivial:
|
| --- x01c.m-orig 2004-01-27 23:46:39.000000000 +0000
| +++ x01c.m 2004-01-27 23:45:31.000000000 +0000
| @@ -103,7 +103,9 @@
| if (status != 0)
| printf("wx=%.3f wy=%.3f dx=%.3f dy=%.3f c=0x%02x str=%s mb=%d
| mod=%0x swin=%d\n", ... wX, wY, dX, dY, keysym, string, button, mod, swin);
| - endif
| + else
| + break;
| + endif
|
| fflush(stdout);
| endwhile
|
| Joao, if you agree with the patch above, I will apply it for the final
| tarball.
Sure, I was only attempting to do what you considered (in another thread) to
be a more intuitive behaviour.
Oddly I can't get the <esc> key status to be reported, but this also happens
in x01c.c.
The code in x01c.c that checks for PLK_Escape is not working, because
plGetCursor returns gin.keysym=0, instead of PLK_Escape. This happens because
in xwin.c:LocateKey() plGinInit() is called *after* PLK_Escape is detected,
setting all the gin structure elements to 0.
I don't know what was the rational for this change, but as it is now we should
clean all x01c like demos of the PLK_Escape code, because being an example
can lead to user confusion.
But please commit your change, the PLK_Escape issue must be corrected
post-release.
| > I suspect that now also occur for the latest octave as well (which I
| > cannot test, of course, but you, Brian, and Rafael have access to it).
|
| I tried plplot-test.sh with Octave 2.1.53 on Debian stable and it fails
| for x03c.m:
|
| error: dimension mismatch in argument x
| error: called from `plline'
| error: evaluating for command near line 44, column 3
| error: called from `x03c' in file `/var/tmp/examples/octave/x03c.m'
| error: evaluating for command near line 4, column 1
|
| This is a 2.1.53-specific problem.
And rises incompatibilities issues, I guess that a lot of users will complain
if this is going to be the default behaviour. You see, where Octave used to
create column vectors, now it creates row vectors! Only a few x??c.c scripts
run OK. I will wait and see.
| We should not bother fixing it
| pre-release, but we should put a note saying that the Octave PLplot
| bindings will not fully work for version > 2.1.50.
|
| > I hope you are also able to sort out the linking problem you found for
| > octave-2.0.17. In my opinion, that is a separate but release-critical
| > mkoctfile issue that should be resolved before release or else
| > octave-2.0.x should be excluded for the release.
|
| I vote for the second option. We should not hold the release just because
| there are version-specific problems with OCtave.
Please notice that the link step was sucesseful when only nn/csa was specified
in the link cmd; when all set of libraries that plplot depends on (freetype,
qhull, etc) was specified, the link failed. So the problem is not mkoctfile
but the libs added to it.
I don't have the time to fix it, feel free to disable octave.
Thanks
Joao
|