plplot-devel

[Plplot-devel] octave now configured with autotools

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

I have just committed some changes that should make octave work at least
for some of the examples.

I tested octave with the installed
prefix/lib/plplot5.1.0/examples/plplot-test.sh script.

The early "p" examples work well, but p7.m and p15.m have problems.
Apparently, the p7.m problem is a long-standing one from comments in
test_octave.sh, but the p15.m problem may be new.

Also, I could not get any of the x??.m examples to work with that script. It
only segfaults.  I presume plplot/test/test_octave.sh (which gets installed
in prefix/lib/plplot5.1.0/examples/) needs to be adjusted to some change that
has been made.

After the early p examples worked, but the x examples segfaulted, I went
back and looked at what was done before.  Apparently there is a sed step to
replace PLPLOT_OCTAVE_PATH with the appropriate value in
plplot_octave_path.m (see below).  I did that step by hand, but it didn't
change either the early p examples working or x examples segfaulting.  So I
think I have gone as far as I can go with this, and it is now up to Joao to
smooth out these rough spots.

Joao, I will be happy to answer any of your questions about what I did to
configure octave with autotools.  But the overall changes that I did to
(mostly) finish octave configuration were to create Makefile.am files (with
appropriate automake information) in the directory tree under
bindings/octave. automake takes care of the rest.  BTW, to do the equivalent
of the sed substitution mentioned above in PLplot, I would suggest simply
renaming plplot_octave_path.m as plplot_octave_path.m.in, putting an @xxx@
variable in the file, and changing configure.ac to configure xxx to whatever
path you want.  Alternatively, you could stick with the sed method for
editing plplot_octave_path.m.in to plplot_octave_path.m that automake will
automatically install. You can stick that rule right in PLplot/Makefile.am,
if you like. It is up to you which method you choose.

Tomorrow, I plan to start working on the static devices.  I don't think that
is going to take long at all. Rafael got them working earlier with the
libtools convenience library approach, and I only plan a slight modification
of his approach so that it doesn't clobber the dynamic driver approach.
After that work is done, I will probably work on java.  That's really a low
priority (since the API is so incomplete), and in fact I was planning to
disable java for the release, but I have essentially run out of things to do
before the release except cheer you guys on to sort out the remaining
rough spots for PLplot.

Currently, those rough spots consist of the -dev tk problem and the above
octave problems. Also, we need lots of testing of this new configuration
system both on Linux and also on other Unix platforms to find any other
rough spots.  But that is about it....

Alan

email: ir...@be...
phone: 250-727-2902	FAX: 250-721-7715
snail-mail:
Dr. Alan W. Irwin
Department of Physics and Astronomy,
University of Victoria, P.O. Box 3055,
Victoria, British Columbia, Canada, V8W 3P6
__________________________

Linux-powered astrophysics
__________________________

Re: [Plplot-devel] octave now configured with autotools

[<jc...@fe...>]

On Thursday 05 December 2002 03:56, Alan W. Irwin wrote:
| I have just committed some changes that should make octave work at
| least for some of the examples.
|
| I tested octave with the installed
| prefix/lib/plplot5.1.0/examples/plplot-test.sh script.
|
| The early "p" examples work well, but p7.m and p15.m have problems.
| Apparently, the p7.m problem is a long-standing one from comments in
| test_octave.sh, but the p15.m problem may be new.
|
| Also, I could not get any of the x??.m examples to work with that
| script. It only segfaults.  I presume plplot/test/test_octave.sh
| (which gets installed in prefix/lib/plplot5.1.0/examples/) needs to
| be adjusted to some change that has been made.
|
| After the early p examples worked, but the x examples segfaulted, I
| went back and looked at what was done before.  Apparently there is a
| sed step to replace PLPLOT_OCTAVE_PATH with the appropriate value in
| plplot_octave_path.m (see below).  I did that step by hand, but it
| didn't change either the early p examples working or x examples
| segfaulting.  So I think I have gone as far as I can go with this,
| and it is now up to Joao to smooth out these rough spots.
|
| Joao, I will be happy to answer any of your questions about what I
| did to configure octave with autotools.

Thanks, Alan, for your work. I will follow the track you lead. Perhaps=20
tomorrow, or in the week-end.

Regarding the examples directory problem: from other packages that use=20
AT, a plain make installs thinks under the distribution tree, with=20
libraries in .libs directories and executables that are linked with the=20
libraries in the .libs dirs, but the real executables becomes hidden=20
(in .libs? no, under .something) and in place of the executable (i.e.=20
with the same name), is put a script that works as a wrapper to the=20
real (hidden) executabl (e.g., seting up a LD_LIBRARY_PATH that founds=20
libs in the .libs dirs).

I'm sure you know all this, but I think that this setup will be fine for=20
us, and avoids the need to install the all package after changing a=20
single line in a source file, in order to run an example.

Joao

|  But the overall changes that
| I did to (mostly) finish octave configuration were to create
| Makefile.am files (with appropriate automake information) in the
| directory tree under bindings/octave. automake takes care of the
| rest.  BTW, to do the equivalent of the sed substitution mentioned
| above in PLplot, I would suggest simply renaming plplot_octave_path.m
| as plplot_octave_path.m.in, putting an @xxx@ variable in the file,
| and changing configure.ac to configure xxx to whatever path you want.
|  Alternatively, you could stick with the sed method for editing
| plplot_octave_path.m.in to plplot_octave_path.m that automake will
| automatically install. You can stick that rule right in
| PLplot/Makefile.am, if you like. It is up to you which method you
| choose.
|
| Tomorrow, I plan to start working on the static devices.  I don't
| think that is going to take long at all. Rafael got them working
| earlier with the libtools convenience library approach, and I only
| plan a slight modification of his approach so that it doesn't clobber
| the dynamic driver approach. After that work is done, I will probably
| work on java.  That's really a low priority (since the API is so
| incomplete), and in fact I was planning to disable java for the
| release, but I have essentially run out of things to do before the
| release except cheer you guys on to sort out the remaining rough
| spots for PLplot.
|
| Currently, those rough spots consist of the -dev tk problem and the
| above octave problems. Also, we need lots of testing of this new
| configuration system both on Linux and also on other Unix platforms
| to find any other rough spots.  But that is about it....
|
| Alan
|
| email: ir...@be...
| phone: 250-727-2902=09FAX: 250-721-7715
| snail-mail:
| Dr. Alan W. Irwin
| Department of Physics and Astronomy,
| University of Victoria, P.O. Box 3055,
| Victoria, British Columbia, Canada, V8W 3P6
| __________________________
|
| Linux-powered astrophysics
| __________________________
|
|
|
| -------------------------------------------------------
| This SF.net email is sponsored by: Microsoft Visual Studio.NET
| comprehensive development tool, built to increase your
| productivity. Try a free online hosted session at:
| http://ads.sourceforge.net/cgi-bin/redirect.pl?micr0003en
| _______________________________________________
| Plplot-devel mailing list
| Plp...@li...
| https://lists.sourceforge.net/lists/listinfo/plplot-devel

Re: [Plplot-devel] octave now configured with autotools

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

On Thu, 5 Dec 2002, [iso-8859-1] Jo=E3o Cardoso wrote:

> Regarding the examples directory problem: from other packages that use
> AT, a plain make installs thinks under the distribution tree, with
> libraries in .libs directories and executables that are linked with the
> libraries in the .libs dirs, but the real executables becomes hidden
> (in .libs? no, under .something) and in place of the executable (i.e.
> with the same name), is put a script that works as a wrapper to the
> real (hidden) executabl (e.g., seting up a LD_LIBRARY_PATH that founds
> libs in the .libs dirs).

>
> I'm sure you know all this, but I think that this setup will be fine for
> us, and avoids the need to install the all package after changing a
> single line in a source file, in order to run an example.

I hadn't actually thought about this before, but I believe you are right.
You might want to flesh out the idea a little more.  For example, there
might be a way to use the machinery of autotools support for tests so that
"make check" compiles all the plplot/examples/x??c programmes in place.
Subsequently, you could run any of those programmes interactively to test
out your change say to libplplot without going through the full install
process (or recompiling the examples since with shared libraries you can
rebuild the library without having to rebuild the application that uses the
library).

Several PLplot generations later we might want to make the tests more
complicated with comparison with test postscript results, but right now suc=
h
complications would make no sense since the tests would often fail due to
platform differences and poor internal precision of PLplot.  So if you
decide to go with the test suite idea, I would keep it really simple and
simply focus on compilation and linking of the uninstalled examples.

Alan

email: ir...@be...
phone: 250-727-2902=09FAX: 250-721-7715
snail-mail:
Dr. Alan W. Irwin
Department of Physics and Astronomy,
University of Victoria, P.O. Box 3055,
Victoria, British Columbia, Canada, V8W 3P6
__________________________

Linux-powered astrophysics
__________________________

Re: [Plplot-devel] octave now configured with autotools

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

On Thu, 5 Dec 2002, Alan W. Irwin wrote:

> You might want to flesh out the idea a little more.  For example, there
> might be a way to use the machinery of autotools support for tests so that
> "make check" compiles all the plplot/examples/x??c programmes in place.

To answer my own comment, I just discovered Autobook section 7.7 which tells
how to do this.  It is so simple that I couldn't resist trying a proof of
concept in examples/c/Makefile.am.  That worked immediately so I have just
committed it.

Joao, please try "make check" and tell me if that satisfies your developer
needs for convenience and speed for the C examples?  If so, then one of us
should follow up with the appropriate stanzas in Makefile.am in
f77, c++, and tk.

After "make check", I tried a number of examples including x08c and x20c,
and the only one I found which fails to work is x14c.  I assume that is
related to the -dev tk problem.

Alan

Re: [Plplot-devel] octave now configured with autotools

[<jc...@fe...>]

On Thursday 05 December 2002 17:56, Alan W. Irwin wrote:
| On Thu, 5 Dec 2002, Alan W. Irwin wrote:
| > You might want to flesh out the idea a little more.  For example,
| > there might be a way to use the machinery of autotools support for
| > tests so that "make check" compiles all the plplot/examples/x??c
| > programmes in place.
|
| To answer my own comment, I just discovered Autobook section 7.7
| which tells how to do this.  It is so simple that I couldn't resist
| trying a proof of concept in examples/c/Makefile.am.  That worked
| immediately so I have just committed it.
|
| Joao, please try "make check" and tell me if that satisfies your
| developer needs for convenience and speed for the C examples?  If so,
| then one of us should follow up with the appropriate stanzas in
| Makefile.am in f77, c++, and tk.
|
| After "make check", I tried a number of examples including x08c and
| x20c, and the only one I found which fails to work is x14c.  I assume
| that is related to the -dev tk problem.
|
| Alan

In examples/c, after I made a "make check", which worked (i.e., compile=20
but not run, see bellow), I made

[jcard@feup] touch x01c.c=20
[jcard@feup] make x01c  =20
source=3D'x01c.c' object=3D'x01c.o' libtool=3Dno \
depfile=3D'.deps/x01c.Po' tmpdepfile=3D'.deps/x01c.TPo' \
depmode=3Dgcc3 /bin/sh ../../depcomp \
gcc -DHAVE_CONFIG_H -I. -I. -I../../include -I../../include =20
-I../../libltdl   -g -O2 -c `test -f x01c.c || echo './'`x01c.c
/bin/sh ../../libtool --mode=3Dlink gcc  -g -O2   -o x01c  x01c.o=20
=2E./../src/libplplot.la=20
gcc -g -O2 -o .libs/x01c x01c.o  ../../src/.libs/libplplot.so -lm -ldl=20
-Wl,--rpath -Wl,/usr/local/lib
creating x01c
[jcard@feup] ./x01c

*** PLPLOT ERROR ***
Can't open drivers/drivers.db
Program aborted

So, this seems sufficient to me; I can have a xterm in examples/c and=20
another in src, modify a source file, "make" to compile it, and in the=20
other xterm make/execute a given example.
Of course, we (:-) only need to solve the drivers.db file problem, and=20
after that make sure that the drivers are found.

Joao.

Re: [Plplot-devel] octave now configured with autotools

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

I believe make x01c is just using ordinary default rules for make so it
isn't finding everything that it needs.  You should be using make check
instead.

First remove everything created by make x01c in examples/c so as not to
confuse issues. Second try make check in that directory. Everything should
work--- at least it did for me!  Of course the first time it will compile
all examples in that directory, but the second and subsequent times it
should only compile the example you touched.

Alan

email: ir...@be...
phone: 250-727-2902=09FAX: 250-721-7715
snail-mail:
Dr. Alan W. Irwin
Department of Physics and Astronomy,
University of Victoria, P.O. Box 3055,
Victoria, British Columbia, Canada, V8W 3P6
__________________________

Linux-powered astrophysics
__________________________

On Thu, 5 Dec 2002, [iso-8859-1] Jo=E3o Cardoso wrote:

> On Thursday 05 December 2002 17:56, Alan W. Irwin wrote:
> | On Thu, 5 Dec 2002, Alan W. Irwin wrote:
> | > You might want to flesh out the idea a little more.  For example,
> | > there might be a way to use the machinery of autotools support for
> | > tests so that "make check" compiles all the plplot/examples/x??c
> | > programmes in place.
> |
> | To answer my own comment, I just discovered Autobook section 7.7
> | which tells how to do this.  It is so simple that I couldn't resist
> | trying a proof of concept in examples/c/Makefile.am.  That worked
> | immediately so I have just committed it.
> |
> | Joao, please try "make check" and tell me if that satisfies your
> | developer needs for convenience and speed for the C examples?  If so,
> | then one of us should follow up with the appropriate stanzas in
> | Makefile.am in f77, c++, and tk.
> |
> | After "make check", I tried a number of examples including x08c and
> | x20c, and the only one I found which fails to work is x14c.  I assume
> | that is related to the -dev tk problem.
> |
> | Alan
>
> In examples/c, after I made a "make check", which worked (i.e., compile
> but not run, see bellow), I made
>
> [jcard@feup] touch x01c.c
> [jcard@feup] make x01c
> source=3D'x01c.c' object=3D'x01c.o' libtool=3Dno \
> depfile=3D'.deps/x01c.Po' tmpdepfile=3D'.deps/x01c.TPo' \
> depmode=3Dgcc3 /bin/sh ../../depcomp \
> gcc -DHAVE_CONFIG_H -I. -I. -I../../include -I../../include
> -I../../libltdl   -g -O2 -c `test -f x01c.c || echo './'`x01c.c
> /bin/sh ../../libtool --mode=3Dlink gcc  -g -O2   -o x01c  x01c.o
> ../../src/libplplot.la
> gcc -g -O2 -o .libs/x01c x01c.o  ../../src/.libs/libplplot.so -lm -ldl
> -Wl,--rpath -Wl,/usr/local/lib
> creating x01c
> [jcard@feup] ./x01c
>
> *** PLPLOT ERROR ***
> Can't open drivers/drivers.db
> Program aborted
>
> So, this seems sufficient to me; I can have a xterm in examples/c and
> another in src, modify a source file, "make" to compile it, and in the
> other xterm make/execute a given example.
> Of course, we (:-) only need to solve the drivers.db file problem, and
> after that make sure that the drivers are found.
>
> Joao.
>
>
>
>
>
> -------------------------------------------------------
> This sf.net email is sponsored by:ThinkGeek
> Welcome to geek heaven.
> http://thinkgeek.com/sf
> _______________________________________________
> Plplot-devel mailing list
> Plp...@li...
> https://lists.sourceforge.net/lists/listinfo/plplot-devel
>

Re: [Plplot-devel] octave now configured with autotools

[<jc...@fe...>]

On Thursday 05 December 2002 20:03, Alan W. Irwin wrote:
| I believe make x01c is just using ordinary default rules for make so
| it isn't finding everything that it needs.  You should be using make
| check instead.
|
| First remove everything created by make x01c in examples/c so as not
| to confuse issues.=20

Nope,

> make distclean
> ./configure --enable-dyndrivers
> make
> make check
=2E..
[jcard@feup] pwd
/home/jcard/plplot
[jcard@feup] examples/c/x01c=20

*** PLPLOT ERROR ***
Can't open drivers/drivers.db

Program aborted

[jcard@feup] cd examples/c
[jcard@feup] ./x01c=20

*** PLPLOT ERROR ***
Can't open drivers/drivers.db

Program aborted

[jcard@feup] cd ../..

[jcard@feup] find . -name drivers.db
=2E/drivers/drivers.db

So it's not working.
Could be that you have an installed  plplot?

Joao

|

Re: [Plplot-devel] octave now configured with autotools

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

On Thu, 5 Dec 2002, [iso-8859-1] Jo=E3o Cardoso wrote:

> So it's not working.
> Could be that you have an installed  plplot?

OOPS.  You are right.  I don't want to remove my installed version because =
I
am right in the middle of something, but now that you have reminded me, I
know you are correct. For example, plcore.c now only looks in one place (th=
e
installed location) for drivers.db, the drivers themselves, and I think the
fonts also.

For now, try with an installed version.  I think it will still link against
the uninstalled libplplot (in case you are testing changes in the core
functions).  In fact after an install try removing everything from the
install location except for drivers, drivers.db and fonts, and I believe
make check will still work.

Ultimately, if you don't want to depend on installed drivers.db, drivers,
and fonts, then you will have to change plcore to search for them in the
uninstalled locations first.

Alan

Re: [Plplot-devel] octave now configured with autotools

[<jc...@fe...>]

On Thursday 05 December 2002 03:56, Alan W. Irwin wrote:
| I have just committed some changes that should make octave work at
| least for some of the examples.
|
| I tested octave with the installed
| prefix/lib/plplot5.1.0/examples/plplot-test.sh script.

I tested in bindings/octave, after a "make install" and changing=20
=2Eoctaverc

| The early "p" examples work well, but p7.m and p15.m have problems.

p7.m had a bug (really shade.m), that I have already correct; p15.m runs=20
OK for me.

| Apparently, the p7.m problem is a long-standing one from comments in
| test_octave.sh, but the p15.m problem may be new.
|
| Also, I could not get any of the x??.m examples to work with that
| script. It only segfaults.  I presume plplot/test/test_octave.sh
| (which gets installed in prefix/lib/plplot5.1.0/examples/) needs to
| be adjusted to some change that has been made.

I don't think so. The problem arises if I plinit(), then plend() and=20
then plinit() again; at the second plinit() I get a seg. fault:

octave:1> plplot_stub=20
octave:2> plsdev("xwin")
octave:3> plinit
octave:4> plend
octave:5> plinit

Plotting Options:
 < 1> panic: Segmentation fault -- stopping myself...
attempting to save variables to `octave-core'...
save to `octave-core' complete
Segmentation fault

This does not happens in my frozen plplot directory (before AT merge),=20
so it is a consequence of the merge.

As a matter of fact, if I add a plinit() at the end of main() in x01c.c,=20
after plend() but before exit(), it has the same behaviour, it's not a=20
octave issue:

[jcard@feup] examples/c/x01c -dev xwin
Plplot library version: 5.1.0

Plotting Options:
Segmentation fault

Making the same modification on my frozen plplot dir just opens another=20
window, then exits without a seg. fault.

Joao

|
| After the early p examples worked, but the x examples segfaulted, I
| went back and looked at what was done before.  Apparently there is a
| sed step to replace PLPLOT_OCTAVE_PATH with the appropriate value in
| plplot_octave_path.m (see below).  I did that step by hand, but it
| didn't change either the early p examples working or x examples
| segfaulting.  So I think I have gone as far as I can go with this,
| and it is now up to Joao to smooth out these rough spots.
|
| Joao, I will be happy to answer any of your questions about what I
| did to configure octave with autotools.  But the overall changes that
| I did to (mostly) finish octave configuration were to create
| Makefile.am files (with appropriate automake information) in the
| directory tree under bindings/octave. automake takes care of the
| rest.  BTW, to do the equivalent of the sed substitution mentioned
| above in PLplot, I would suggest simply renaming plplot_octave_path.m
| as plplot_octave_path.m.in, putting an @xxx@ variable in the file,
| and changing configure.ac to configure xxx to whatever path you want.
|  Alternatively, you could stick with the sed method for editing
| plplot_octave_path.m.in to plplot_octave_path.m that automake will
| automatically install. You can stick that rule right in
| PLplot/Makefile.am, if you like. It is up to you which method you
| choose.
|
| Tomorrow, I plan to start working on the static devices.  I don't
| think that is going to take long at all. Rafael got them working
| earlier with the libtools convenience library approach, and I only
| plan a slight modification of his approach so that it doesn't clobber
| the dynamic driver approach. After that work is done, I will probably
| work on java.  That's really a low priority (since the API is so
| incomplete), and in fact I was planning to disable java for the
| release, but I have essentially run out of things to do before the
| release except cheer you guys on to sort out the remaining rough
| spots for PLplot.
|
| Currently, those rough spots consist of the -dev tk problem and the
| above octave problems. Also, we need lots of testing of this new
| configuration system both on Linux and also on other Unix platforms
| to find any other rough spots.  But that is about it....
|
| Alan
|
| email: ir...@be...
| phone: 250-727-2902=09FAX: 250-721-7715
| snail-mail:
| Dr. Alan W. Irwin
| Department of Physics and Astronomy,
| University of Victoria, P.O. Box 3055,
| Victoria, British Columbia, Canada, V8W 3P6
| __________________________
|
| Linux-powered astrophysics
| __________________________
|
|
|
| -------------------------------------------------------
| This SF.net email is sponsored by: Microsoft Visual Studio.NET
| comprehensive development tool, built to increase your
| productivity. Try a free online hosted session at:
| http://ads.sourceforge.net/cgi-bin/redirect.pl?micr0003en
| _______________________________________________
| Plplot-devel mailing list
| Plp...@li...
| https://lists.sourceforge.net/lists/listinfo/plplot-devel