[Plplot-devel] Poor RC1 test results on Debian woody

[Alan W. I. <ir...@be...>]

I was just able to find some time to test RC1 on my Debian woody system.
Sadly, it does not give a well-tested impression.  I was really relying on
the rest of you to do some extensive checks for tcl/tk and octave, but it
appears those have not been done in even a superficial manner.  For a number
of good reasons I really am scaling back my testing role to superficial
checks that anybody can do on their own system, and somebody else here has
to take up the slack if we are going to have a quality product.

That said, I have fixed the obvious stuff.  If in addition, Rafael or Joao
can fix the p16 octave example (and anything else that shows up for a
complete octave test), then I think Rafael should go ahead with RC2 early
tomorrow (Wednesday).

Here is what I did:

./configure --prefix=/usr/local/plplot_at --enable-java --disable-static
--with-prebuiltdoc >& configure.out

make; make install

A superficial check of the various kinds of installed documentation went well.

plplot-test.sh built psc and plmeta files with no obvious problems (other than
an obvious pladv ordering bug in x21c which I have just fixed).

I was able to do some minimal tcl/tk testing as well (although Maurice and
Geoffrey should be responsible for most of that testing).

*** Simple wish and tclsh tests following the directions in README.tcldemos
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 any
longer).  To solve this I have introduced LIBRARY_VERSION = 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 keeping LIBRARY_VERSION
consistent with SOVERSION?  That's important because otherwise we have a
tcl/tk maintenance issue that will come back to haunt us every time
SOVERSION is changed.  I did look for any variable that was already set to
the numerical library suffix, but I couldn't find any.

*** There was one nasty RC1 failure for the combination of octave and png
device.

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

followed by a whole bunch of garbage output.

This should have easily been found with even superficial octave testing unless
there is some platform difference between me on the one hand and Joao and
Rafael on the other.
Rafael and Joao, what are your results for
./plplot-test.sh --device=png
( or ./plplot-test.sh --front-end=octave --device=png) for RC1.

If they verify this problem I suspect it is due to some recent change to p16
that does something out of order since that example worked before for me
with png.

Warnings.

I did notice some warnings along the way.  configure still outputs the
following:

configure: WARNING: itclDecls.h: present but cannot be compiled
configure: WARNING: itclDecls.h: check for missing prerequisite headers?
configure: WARNING: itclDecls.h: proceeding with the preprocessor's result
configure: WARNING:     ## ------------------------------------ ##
configure: WARNING:     ## Report this to bug...@gn.... ##
configure: WARNING:     ## ------------------------------------ ##

This is a symptom of the poor state of our unix/linux tcl/tk configuration.
I wanted to replace that configuration, but I didn't feel right attempting
that since I am no tcl/tk expert.  I noticed, however, that the tclConfig.sh
that comes with tcl8.3 is deprecated and it is not immediately obvious how
to use it with autotools. Its suggested replacement, tcl.m4 is supposed to
fit in with autotools, but it needs an m4/autoconf expert to look at it to
see if it is consistent with the current generation of autotools.  (I
suspect it is not from all the name clashes I see.) Whether that is a post
or pre-release issue really depends on how well our current tcl/tk
configuration works cross-platform. Maurice and Geoffrey, does the current
tcl/tk configuration work okay on all Linux/Unix platforms accessible to
you?

Here are some compiler warnings that I noticed:

* plplotcmodule_double.c:684: warning: PySequence_Fast_GET_ITEM' redefined
//usr/include/python2.1/abstract.h:902: warning: this is the location of the
previous definition

I believe this is a swig issue that we have no control over (except to try
later swig versions post-release and send in bug reports to swig if the
issue is still there).

* /usr/bin/g++ -c -fPIC -I/usr/include/octave-2.1.35
-I/usr/include/octave-2.1.35/
octave -I/usr/include -mieee-fp -fno-implicit-templates -O2 -I.
plplot_octave.cc
 -o plplot_octave.o
 In file included from /usr/include/octave-2.1.35/octave/oct.h:31,
                  from plplot_octave.cc:11:
		  /usr/include/octave-2.1.35/octave/config.h:148: warning:
HAVE_ISINF' redefined
../../config.h:36: warning: this is the location of the previous definition
/usr/include/octave-2.1.35/octave/config.h:151: warning: HAVE_ISNAN' redefined
../../config.h:39: warning: this is the location of the previous definition
/usr/include/octave-2.1.35/octave/config.h:288: warning: HAVE_FINITE'
redefined
../../config.h:24: warning: this is the location of the previous definition
/usr/include/octave-2.1.35/octave/config.h:474: warning: HAVE_USLEEP'
redefined
../../config.h:101: warning: this is the location of the previous definition

Rafael or Joao may want to deal with these warnings before the release.

* Whole host of tk-x-plat/plplotter.c warnings.  IIRC, Vince said those
disappear if you use a version of tcl that is later than 8.3.
Alternatively, these symptoms may be a result of our iffy tcl/tk
configuration.

That (and a quick check of RC2 for the above issues) is about all the
testing I have time for before the release so again, somebody else is going
to have to step forward if you want a quality release.

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 package
(plplot.org).

__________________________

Linux-powered Science
__________________________