Menu
▾
▴
Re: [Plplot-devel] Poor RC1 test results on Debian woody
|
From: Alan W. I. <ir...@be...> - 2003-04-16 02:00:23
|
On Tue, 15 Apr 2003, Rafael Laboissiere wrote:
> * Alan W. Irwin <ir...@be...> [2003-04-15 14:03]:
>
> > *** Simple wish and tclsh tests following the directions in README.tcld=
emos
> > and README.tkdemos don't work in RC1 because there is a problem for
> > bindings/*/pkgIndex.tcl.in. Those scripts assume the library suffix is=
the
> > same as VERSION (true at the time I introduced those scripts, but not a=
ny
> > longer). To solve this I have introduced LIBRARY_VERSION =3D numerical
> > suffix of the library in configure.ac which must be adjusted to be
> > consistent with SOVERSION. Rafael, can you find an automatic way of kee=
ping
> > LIBRARY_VERSION consistent with SOVERSION? That's important because
> > otherwise we have a tcl/tk maintenance issue that will come back to hau=
nt
> > us every time SOVERSION is changed. I did look for any variable that w=
as
> > already set to the numerical library suffix, but I couldn't find any.
>
> I dislike this solution for two reason reason. the first, as you mention
> above, is the maintenance burden that it introduces. The second (and mor=
e
> important) reason is that your assumption of library suffix generation do=
es
> not work cross-platform. If I remember correctly, in HPUX (for instance)
> when SOVERSION=3D8:0:3, then the library would be something like
> libplplot.sl.8.3. There are other system with even more exotic rule and =
the
> assumption that it will be libplplot.5.3.0 is far from being the rule.
I also agree with your second point (as you will see from my configure.ac
notes). However, this is just a quick fix to at least get plplot to work
under wish and tclsh for Linux, and the quick fix might even work for some
unices such as solaris. We obviously need a more fundamental solution here=
,
but that needs a tcl/tk expert to step in.
The simple problem being solved here is that wish and tclsh need to know th=
e
_complete_ PLplot library names so they can load them. Maurice, I wonder i=
f
you could use the libtool results $prefix/lib/lib*.la to find the complete
library names that wish and tclsh need? For example, the libplplotd versio=
n
of this file installed on my system has the following line:
library_names=3D'libplplotd.so.5.3.0 libplplotd.so.5 libplplotd.so' and
similarly for libplplottcltkd
library_names=3D'libplplottcltkd.so.5.3.0 libplplottcltkd.so.5 libplplottcl=
tkd.so'
If solaris lib*.la files have something similar, Maurice or Geoffry might b=
e
able to come up with a good cross-platform way to deal with this issue in
pkgIndex.tcl(.in) using the tcl equivalent of grep and cut to obtain the
first library name mentioned in the list. Because Maurice is going on
vacation, this might be worth delaying the release for a few days.
>
> > Rafael and Joao, what are your results for
> > ./plplot-test.sh --device=3Dpng
> > ( or ./plplot-test.sh --front-end=3Doctave --device=3Dpng) for RC1.
>
> $ ./plplot-test.sh --front-end=3Doctave --device=3Dpng
> [removed warnings about automatic_replot]Opened p1.png
> Opened p2.png
> Opened p3.png
> Opened p4.png
> Opened p5.png
> Opened p6.png
> Opened p7.png
> Opened p8.png
> Opened p9.png
> Opened p13.png
> Opened p15.png
> Opened p16.png
> Plplot library version: 5.2.0.cvs.20030415
> Opened x01o.png
> [removed mouse click message]
> Opened x02o.png
> Opened x03o.png
> Opened x04o.png
> Opened x05o.png
> Opened x06o.png
> Opened x07o.png
> Opened x08o.png
> Opened x09o.png
> Opened x10o.png
> Opened x11o.png
> Opened x12o.png
> Opened x13o.png
> Opened x15o.png
> Opened x16o.png
> Opened x18o.png
Well, I certainly do _not_ get those results on Debian woody. It appears w=
e
have now found two bugs. The result you have above is completely incorrect
because there is no familying. I have no idea why this bug shows up for
your system. Does this occur just for octave? Could you give the
=2E/plplot-test.sh --front-end=3Dc --device=3Dpng results as well? There s=
hould
be many pages for most of the x examples, yet it appears you only have one.
If the lack of familying persists for the C front end for device=3Dpng, the=
n
please have a look at the plplot-test.sh script to see what is going on. If
it only occurs for octave, then somehow octave is not getting the familying
parameter passed to it correctly on your Debian testing version platform,
while that works fine on Debian woody. Very strange indeed!
Here is the complete Debian woody result. It _does_ have the proper
familying (note the different style of file names and the multiple pages fo=
r
some examples), but it also dies horribly during p16 (Rafael would probably
see this as well if familying were being correctly turned on for his case).
=2E/plplot-test.sh --front-end=3Doctave --device=3Dpng
warning: It is recommended that you set 'automatic_replot=3D1'
in your ~/.octaverc file.
warning: in /usr/local/plplot_at/share/plplot_octave/figure.m near line 59,=
column 7:
>>> warning ("It is recommended that you set 'automatic_replot=3D1' \n\t i=
n your ~/.octaverc file.");
Opened p1.png.01
Opened p2.png.01
Opened p3.png.01
Opened p4.png.01
Opened p5.png.01
Opened p6.png.01
Opened p7.png.01
Opened p7.png.02
Opened p8.png.01
Opened p8.png.02
Opened p9.png.01
Opened p13.png.01
Opened p15.png.01
Opened p15.png.02
Opened p16.png.01
*** PLPLOT WARNING ***
plsfam: Must be called before plinit.
*** PLPLOT WARNING ***
plsdev: Must be called before plinit.
Plplot library version: 5.2.1.rc1
(AWI comment added. I think the above two warning messages are the key to =
the
p16 problem).
Plotting Options:
< 1> xwin X-Window (Xlib)
< 2> tk Tcl/TK Window
< 3> xterm Xterm Window
< 4> tekt Tektronix Terminal (4010)
< 5> tek4107t Tektronix Terminal (4105/4107)
< 6> mskermit MS-Kermit emulator
< 7> versaterm Versaterm vt100/tek emulator
< 8> vlt VLT vt100/tek emulator
< 9> conex Conex vt320/tek emulator
<10> dg300 DG300 Terminal
<11> plmeta PLplot Native Meta-File
<12> tekf Tektronix File (4010)
<13> tek4107f Tektronix File (4105/4107)
<14> ps PostScript File (monochrome)
<15> psc PostScript File (color)
<16> xfig Fig file
<17> ljiip LaserJet IIp/deskjet compressed graphics
<18> ljii LaserJet II Bitmap File (150 dpi)
<19> hp7470 HP 7470 Plotter File (HPGL Cartridge, Small Plotter)
<20> hp7580 HP 7580 Plotter File (Large Plotter)
<21> lj_hpgl HP Laserjet III, HPGL emulation mode
<22> imp Impress File
<23> pbm PDB (PPM) Driver
<24> png PNG file
<25> jpeg JPEG file
<26> pstex Combined Postscript/LaTeX files
<27> null Null device
<28> ntk New tk driver
<29> cgm CGM file
<30> mem User-supplied memory device
<31> tkwin New tk driver
Enter device number or keyword:
Invalid device: =AC=E1=FF=BF=D3&=CB@=D0=C67
This message is repeated many times before....
*** PLPLOT ERROR ***
plSelectDev: Too many tries.
Program aborted
Invalid device: =AC=E1=FF=BF=D3&=CB@=D0=C6
Because of the two messages about plinit, I think there is an ordering prob=
lem
in p16 which has been recently introduced. Joao, can you replicate this
bug or has familying disappeared in your environment as well?
> It seems that most of the problems that you found are quite minor or even
> negligeable for now (besides that LIBRARY_VERSION bizareness). I think t=
hat
> we are pretty close to a "quality release".
Well, I hope you are right. Certainly, I think only modest changes are
required to get to a quality release. But "the devil is in the details".
The biggest problem known at this time is the bizarre lack of familying on
Rafael's Debian platform. That really kills all png results for at least
octave (if not the other front ends as well depending on the results of
Rafael's additional familying tests.) There is also what I hope is a minor
ordering issue with p16 for octave.
Rafael, I now think you should hold off on RC2 until at least these two
issues are resolved, and furthermore we may want to delay the final release
by a few days depending on what Maurice says about the wish/tclsh library
finding problem.
Alan
__________________________
Alan W. Irwin
email: ir...@be...
phone: 250-727-2902
Astronomical research affiliation with Department of Physics and Astronomy,
University of Victoria (astrowww.phys.uvic.ca).
Programming affiliation with the PLplot scientific plotting software packag=
e
(plplot.org).
__________________________
Linux-powered Science
__________________________
|
