plplot-devel — Technical discussion about PLplot development

Re: [Plplot-devel] need help with atexit error

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

On Tue, 29 Jan 2002, Joao Cardoso wrote:

> On Tuesday 29 January 2002 7:24 pm, Alan W. Irwin wrote:
> > More information (hot off the presses).  I hope you read all your mail
> > before responding to the previous messages in this thread.
> >
> >
> > Have you cross-platform guys tried all the examples?  I wonder how many
> > other platforms demand the -ltclmatrixd link line?  I wonder how Debian
> > woody works fine without requiring it?

> > [...]If I add -ltclmatrixd to the link lines in the examples Makefile,
> > all C and Tk build and run without problems.  So problem solved, and
> > thanks for listening as I worked through this.
>
> I don't understand, as I get -ltclmatrixd in the link automatically:
>
> [jcard@feup] make tkdemos
> gcc -c          -O    -I. -I/usr/lib/java//include xtk01.c
> gcc   xtk01.o \
>        -L. -lplplotd -ltclmatrixd -o xtk01 -litk3.2 -ltk8.3 -litcl3.2
> -ltcl8.3 -L/usr/X11R6/lib -lX11 -ldl -lm -lg2c -Wl,-rpath
> -Wl,/home/jcard/plplot-devel/tmp

I wasn't clear.  Everything is fine in the tmp location.  It is the
installed location ($prefix/lib/plplot5.1.0/examples/c, for example) where
the make cdemos fails on RH 7.1 (but not on Debian/woody).  The tmp Makefile
has the -ltclmatrixd, but the installed examples one does not, and I am
about to fix that since it does make a difference for RH 7.1.

>
> even if I remove the tcllibrary, it links OK;
> [jcard@feup] gcc   xtk01.o -L. -lplplotd  -o xtk01 -litk3.2 -ltk8.3 -litcl3.2
> -ltcl8.3 -L/usr/X11R6/lib -lX11 -ldl -lm -lg2c -Wl,-rpath
> -Wl,/home/jcard/plplot-devel/tmp

Yep, that works on Debian/woody as well, but not RH 7.1.

>
> But, with a standard make, "xtk01"f ails:
>
> [jcard@feup] ./xtk01
> % myplot
> Segmentation fault
>
> [jcard@feup] ./xtk01
> % myplot 1
>
> *** PLPLOT ERROR ***
> plcol0: Please call plinit first, aborting operation
> ...
> and so on.
>
> The README.tkdemos does not provides accurate instrutions on runing this
> component of the demos, so I don't know if this is the correct way--I had to
> read the xtk01.c code.

Actually, the test_tk.sh script I shared with the list before is an excellent
guide, but it needs a slight generalization so it can work from both
the tmp and installed locations.

Here is the modified script (it is set up for the installed examples
location, but set the first 4 export lines to . to work
from the tmp location.)
***********************
#!/bin/sh
#test suite of all interactive tk stuff that cannot be done by file.
#tk driver (x14c), tcldemos, tkdemos.
export cdir=c
export tcldir=tcl
export tkdir=tk
export plplotbindir=/usr/local/plplot/bin
$cdir/x14c
$cdir/x17c -dev tk
$tkdir/xtk01 -f $tkdir/tk01
$tkdir/xtk02 -f $tkdir/tk02
$plplotbindir/plserver -f $tkdir/tk03
$tkdir/xtk04 -f $tkdir/tk04
cd $tcldir
$plplotbindir/plserver <<EOF
plstdwin .
plxframe .plw
pack append . .plw {left expand fill}
source plgrid.tcl
proc 1 {} "plgrid .plw.plwin"
1
exit
EOF
cd ../$tkdir
pwd
$plplotbindir/plserver <<EOF
source tkdemos.tcl
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
exit
EOF

***********************

Also, you must execute the following before this script if you want to
exercise the installed examples.

pushd /usr/local/plplot/lib/plplot5.1.0/examples/
cd c; make cdemos >&! make_cdemos.out; cd ..
cd f77 ; make fdemos > & ! make_fdemos.out ; cd ..
cd c++ ; make cxxdemos > & ! make_cxxdemos.out ; cd ..
cd tk ; make tkdemos > & ! make_tkdemos.out ; cd ..

Let me know whether the above works (once I make the commit of my
configuration change).

>
> I have further tested:
>
> cdemos-OK
> fdemos-OK
> cxxdemos-OK
> jdemos-OK
> pythondemos-OK
> tcldemos-OK
> octave demos-OK
> tkdemos-half OK, the "start plserver, source tkdemos etc" worked, the xtk??
> demos didn't.

Thanks, Joao.  Please try this from the installed examples location as well,
once I get the configuration change out so the demo Makefile there is
correct for RH 7.1.

Alan

Re: [Plplot-devel] need help with atexit error

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

On Tuesday 29 January 2002 7:24 pm, Alan W. Irwin wrote:
> More information (hot off the presses).  I hope you read all your mail
> before responding to the previous messages in this thread.
>
>
> Have you cross-platform guys tried all the examples?  I wonder how many
> other platforms demand the -ltclmatrixd link line?  I wonder how Debian
> woody works fine without requiring it?

atexit() is in libc:

[jcard@feup] nm -p /lib/libc.so.6 | fgrep atexit
001161ac ? __elf_set___libc_atexit_element__cleanup__
00118aa0 b added_atexit_handler.186
001161ac a __start___libc_atexit
001161b0 a __stop___libc_atexit
0002f9a0 T __cxa_atexit
0002f960 T atexit
        ^^^
here it is defined as externaly visible.

> If I add -ltclmatrixd to the link lines in the examples Makefile,
> all C and Tk build and run without problems.  So problem solved, and
> thanks for listening as I worked through this.

I don't understand, as I get -ltclmatrixd in the link automatically:

[jcard@feup] make tkdemos
gcc -c          -O    -I. -I/usr/lib/java//include xtk01.c
gcc   xtk01.o \
       -L. -lplplotd -ltclmatrixd -o xtk01 -litk3.2 -ltk8.3 -litcl3.2=20
-ltcl8.3 -L/usr/X11R6/lib -lX11 -ldl -lm -lg2c -Wl,-rpath=20
-Wl,/home/jcard/plplot-devel/tmp

even if I remove the tcllibrary, it links OK;
[jcard@feup] gcc   xtk01.o -L. -lplplotd  -o xtk01 -litk3.2 -ltk8.3 -litc=
l3.2=20
-ltcl8.3 -L/usr/X11R6/lib -lX11 -ldl -lm -lg2c -Wl,-rpath=20
-Wl,/home/jcard/plplot-devel/tmp

But, with a standard make, "xtk01"f ails:

[jcard@feup] ./xtk01=20
% myplot
Segmentation fault

[jcard@feup] ./xtk01=20
% myplot 1

*** PLPLOT ERROR ***
plcol0: Please call plinit first, aborting operation
=2E..
and so on.

The README.tkdemos does not provides accurate instrutions on runing this=20
component of the demos, so I don't know if this is the correct way--I had=
 to=20
read the xtk01.c code.

I have further tested:

cdemos-OK
fdemos-OK
cxxdemos-OK
jdemos-OK
pythondemos-OK
tcldemos-OK
octave demos-OK
tkdemos-half OK, the "start plserver, source tkdemos etc" worked, the xtk=
??=20
demos didn't.

Joao

Re: [Plplot-devel] need help with atexit error

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

More information (hot off the presses).  I hope you read all your mail
before responding to the previous messages in this thread.

If I add -ltclmatrixd to the link lines in the examples Makefile,
all C and Tk build and run without problems.  So problem solved, and
thanks for listening as I worked through this.

Have you cross-platform guys tried all the examples?  I wonder how many
other platforms demand the -ltclmatrixd link line?  I wonder how Debian
woody works fine without requiring 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
__________________________

On Tue, 29 Jan 2002, Alan W. Irwin wrote:

> I am now discovering more than I want to about atexit.  I was wrong below,
> atexit is not obscure at all, but that leads to new questions.
>
> It appears in tek.c
> (and commented out in gnome.c).  Here are the tek.c lines found by grep.
>
> tek.c:static void tty_atexit    (void);
> tek.c:static void tty_atexit    (void) {}
> tek.c:    if (atexit(tty_atexit))
> tek.c:  fprintf(stderr, "Unable to set up atexit handler.\n");
> tek.c:tty_atexit(void)                  /* exit handler */
>
> atexit (which I am sure you knew, but I didn't) is a standard K&R function
> (see p. 253 of the K&R book) defined like
>
> int atexit(void (*fcn)(void))
>
> from p. 119 the & operator is not needed on the function name in argument
> lists so
> if (atexit(tty_atexit)) above seems fine to me.
>
> Also, a simple test programme consisting of
>
> #include <stdio.h>
>
> main()
> {
>
>       printf("Hello, World!\n");
>       atexit();
> }
>
> compiles and links fine with the command
>
> gcc test.c
>
> on both my Debian ***and RH 7.1 systems***.  (Of course it segfaults upon
> execution after printing out the Hello World! message since the argument is
> incorrect on atexit.) So as expected from the K&R book, atexit is just part
> of the standard library on both systems, and nothing special has to be done
> to link against it.
>
> So I am completely stumped why I am getting a linking error concerning
> atexit when I attempt to link either our C or tk examples, but not
> when I link our C++ or fortran examples.
>
> 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
> __________________________
>
> On Tue, 29 Jan 2002, Alan W. Irwin wrote:
>
> > On my RedHat 7.1 system I cannot get either the C or tk examples to build
> >
> > a typical error message is
> >
> > gcc -c          -O    -I/usr/include/plplot/..  -I. -I/usr/include/gtk-1.2
> > -I/us
> > r/include/glib-1.2 -I/usr/lib/glib/include -I/usr/X11R6/include
> > -I/usr/include -
> > I/usr/lib/gnome-libs/include xtk01.c
> > gcc   xtk01.o \
> >        -L/usr/lib -lplplotd -o xtk01 -litk3.1 -ltk8.3 -litcl3.1 -ltcl8.3
> > -L/usr/
> > X11R6/lib -lX11 -ldl -lm -lg2c -Wl,-rpath,/usr/lib
> > /usr/lib/libtclmatrixd.so.5: undefined reference to atexit'
> > collect2: ld returned 1 exit status
> > make: *** [xtk01] Error 1
> >
> > Interestingly, the fortran and c++ examples build and execute fine.
> >
> > I tried replacing gcc with g77 in the above, but it made no difference.
> >
> > I suspect there is some obscure library containing atexit that I am missing
> > on the RH 7.1 system, but which library is it?  Once that is identified, I
> > can install the library and proceed.  Another possibility is that
> > libtclmatrixd is not linked properly, but again I need to know the
> > name of the library which contains atexit.
> >
> > Any help in identifying the library containing atexit would be much
> > appreciated.
> >
> > 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
> > __________________________
> >
> >
> > _______________________________________________
> > Plplot-devel mailing list
> > Plp...@li...
> > https://lists.sourceforge.net/lists/listinfo/plplot-devel
> >
>
>
> _______________________________________________
> Plplot-devel mailing list
> Plp...@li...
> https://lists.sourceforge.net/lists/listinfo/plplot-devel
>

Re: [Plplot-devel] need help with atexit error

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

I am now discovering more than I want to about atexit.  I was wrong below,
atexit is not obscure at all, but that leads to new questions.

It appears in tek.c
(and commented out in gnome.c).  Here are the tek.c lines found by grep.

tek.c:static void tty_atexit    (void);
tek.c:static void tty_atexit    (void) {}
tek.c:    if (atexit(tty_atexit))
tek.c:  fprintf(stderr, "Unable to set up atexit handler.\n");
tek.c:tty_atexit(void)                  /* exit handler */

atexit (which I am sure you knew, but I didn't) is a standard K&R function
(see p. 253 of the K&R book) defined like

int atexit(void (*fcn)(void))

from p. 119 the & operator is not needed on the function name in argument
lists so
if (atexit(tty_atexit)) above seems fine to me.

Also, a simple test programme consisting of

#include <stdio.h>

main()
{

      printf("Hello, World!\n");
      atexit();
}

compiles and links fine with the command

gcc test.c

on both my Debian ***and RH 7.1 systems***.  (Of course it segfaults upon
execution after printing out the Hello World! message since the argument is
incorrect on atexit.) So as expected from the K&R book, atexit is just part
of the standard library on both systems, and nothing special has to be done
to link against it.

So I am completely stumped why I am getting a linking error concerning
atexit when I attempt to link either our C or tk examples, but not
when I link our C++ or fortran examples.

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
__________________________

On Tue, 29 Jan 2002, Alan W. Irwin wrote:

> On my RedHat 7.1 system I cannot get either the C or tk examples to build
>
> a typical error message is
>
> gcc -c          -O    -I/usr/include/plplot/..  -I. -I/usr/include/gtk-1.2
> -I/us
> r/include/glib-1.2 -I/usr/lib/glib/include -I/usr/X11R6/include
> -I/usr/include -
> I/usr/lib/gnome-libs/include xtk01.c
> gcc   xtk01.o \
>        -L/usr/lib -lplplotd -o xtk01 -litk3.1 -ltk8.3 -litcl3.1 -ltcl8.3
> -L/usr/
> X11R6/lib -lX11 -ldl -lm -lg2c -Wl,-rpath,/usr/lib
> /usr/lib/libtclmatrixd.so.5: undefined reference to atexit'
> collect2: ld returned 1 exit status
> make: *** [xtk01] Error 1
>
> Interestingly, the fortran and c++ examples build and execute fine.
>
> I tried replacing gcc with g77 in the above, but it made no difference.
>
> I suspect there is some obscure library containing atexit that I am missing
> on the RH 7.1 system, but which library is it?  Once that is identified, I
> can install the library and proceed.  Another possibility is that
> libtclmatrixd is not linked properly, but again I need to know the
> name of the library which contains atexit.
>
> Any help in identifying the library containing atexit would be much
> appreciated.
>
> 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
> __________________________
>
>
> _______________________________________________
> Plplot-devel mailing list
> Plp...@li...
> https://lists.sourceforge.net/lists/listinfo/plplot-devel
>

[Plplot-devel] need help with atexit error

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

On my RedHat 7.1 system I cannot get either the C or tk examples to build

a typical error message is

gcc -c          -O    -I/usr/include/plplot/..  -I. -I/usr/include/gtk-1.2
-I/us
r/include/glib-1.2 -I/usr/lib/glib/include -I/usr/X11R6/include
-I/usr/include -
I/usr/lib/gnome-libs/include xtk01.c
gcc   xtk01.o \
       -L/usr/lib -lplplotd -o xtk01 -litk3.1 -ltk8.3 -litcl3.1 -ltcl8.3
-L/usr/
X11R6/lib -lX11 -ldl -lm -lg2c -Wl,-rpath,/usr/lib
/usr/lib/libtclmatrixd.so.5: undefined reference to atexit'
collect2: ld returned 1 exit status
make: *** [xtk01] Error 1

Interestingly, the fortran and c++ examples build and execute fine.

I tried replacing gcc with g77 in the above, but it made no difference.

I suspect there is some obscure library containing atexit that I am missing
on the RH 7.1 system, but which library is it?  Once that is identified, I
can install the library and proceed.  Another possibility is that
libtclmatrixd is not linked properly, but again I need to know the
name of the library which contains atexit.

Any help in identifying the library containing atexit would be much
appreciated.

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] a few glitches in the test process. Also, a lot of test progress

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

On Tuesday 29 January 2002 3:08 am, Jo=E3o Cardoso wrote:
> On Monday 28 January 2002 21:47, Geoffrey Furnish wrote:
> | Alan W. Irwin writes:
> |  > Geoffrey, Maurice, and Joao: you have all done some
> |  > cross-platform testing recently.  In the release notes can I
> |  > describe that testing as Solaris/cc, OSF1/Dec Alpha, and SuSe
> |  > 7.1?  Please let me know if I have the qualifiers right and/or
> |  > whether I need more qualifiers to describe those platforms you
> |  > tested on.
> |
> | I have built a clean checkout today on both SuSE-7.2 and Solaris
> | 5.7.
>
> Also on suse-7.0 and suse-7.2 (both with with static and dyndrivers)
> and on SCO-3.2v5.0.4 (with-gcc, gmake, no shared libs, no dyndrivers).
>
> I'm sorry to report that on a "OSF1 V4.0 alpha" I was not able to
> complete the build, although I tried several  configuration options
> (with/without gcc, with gmake, without-shlib,...):

Correction!

I don't know what happened yesterday night! I'm able to build and run the=
=20
current plplot cvs version on an "OSF1 V4.0 alpha", using either the nati=
ve C=20
compiler or gcc 3.0.3, without shared libs and without dyndrivers.

Joao

RE: [Plplot-devel] Deadline for 5.1.0 is Monday 8 AM PST = 16:00 UT

[Maurice L. <mj...@ga...>]

Maurice LeBrun writes:
 > If you start up another plserver & x01c, the main window names become
 > "plserver #2" and "x01c #2", etc.

Hmm.. there is a problem with the latter right now, although there's a
workaround.  If I do:

$ plserver -name plserver -client_name x01c
$ x01c -dev tk -server_name plserver

in one pair of windows, and then repeat in a second pair using the #2 main
window name change, I get:

$ plserver -name "plserver #2" -client_name "x01c #2"
$ x01c -dev tk -server_name "plserver #2"
Plplot library version: 5.1.0
TCL command "plclient_link_init" failed:
         no application named "{plserver #2}"
TCL command "plclient_link_end" failed:
         can't read "plwindow": no such variable
Program aborted

So some problem with list-ification in the tk driver.  The workaround is:

$ plserver -name plserver2 -client_name "x01c #2"
$ x01c -dev tk -server_name plserver2

which is nicer anyway.

-- 
Maurice LeBrun    mj...@ga...

Re: [Plplot-devel] a few glitches in the test process. Also, a lot of test progress

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

Thanks very much for the complete response I got from everybody on the
cross-platform testing.  It will make a most impressive list for the release
announcement.

The current release status is I am tired and heading off to sleep....;-)
Tomorrow, I want to make complete tests of the recent changes to the install
location of the drivers directory on my Debian woody system, and then
continue with my RH 7.1 work (where I am making good progress on build and
install questions, but no testing has been done as yet.) I hope to finish
both the Debian and RH systems tomorrow, and perhaps even start with the
SF compile farm solaris/cc system, which is the final system I have to
deal with.

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
__________________________

On Tue, 29 Jan 2002, Maurice LeBrun wrote:

> Alan W. Irwin writes:
>  > Geoffrey, Maurice, and Joao: you have all done some cross-platform testing
>  > recently.  In the release notes can I describe that testing as Solaris/cc,
>  > OSF1/Dec Alpha, and SuSe 7.1?  Please let me know if I have the qualifiers
>  > right and/or whether I need more qualifiers to describe those platforms you
>  > tested on.
>
> I tried last week on the following systems with no problems:
>
> zion$ uname -a
> OSF1 zion.naka.jaeri.go.jp V4.0 564 alpha
>
> naka-spf$ uname -a
> IRIX64 naka-spf 6.5 04131233 IP35 mips
>
> I'll give the latest another try soon to make sure nothing odd's come up.
>
> --
> Maurice LeBrun    mj...@ga...
>
> _______________________________________________
> Plplot-devel mailing list
> Plp...@li...
> https://lists.sourceforge.net/lists/listinfo/plplot-devel
>

RE: [Plplot-devel] Deadline for 5.1.0 is Monday 8 AM PST = 16:00 UT

[Maurice L. <mj...@ga...>]

Geoffrey Furnish writes:
 > Is there an easy way to run an app against a seperately started
 > plserver?  I seem to remember traffic on this list about this, but I
 > wasn't paying enough attention then to remember how to do it now.

Yeah, Joao tried it and couldn't get it to work, and I wasn't much help at the
time.  It certainly did work at one time, else I wouldn't have documented it
in the code (scenario #3 in tk.c:init_server()).  My first few tries were
unsuccessful, but then..  I got it to work, yeah!  (my wife's asleep & I can't
get the dogs to cheer, oh well. :)

OK, so here's the scoop.  In one window:

$ plserver -name plserver -client_name x01c

and then in another:

$ x01c -dev tk -server_name plserver

... ta-da!

I remember doing considerable debugging in this mode b/c I had access to the
Tcl shell in window #1, so I could find out lots of nice info that way.

If you start up another plserver & x01c, the main window names become
"plserver #2" and "x01c #2", etc.

-- 
Maurice LeBrun    mj...@ga...

Re: [Plplot-devel] Install location for the drivers directory?

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

On Tue, 29 Jan 2002, [iso-8859-1] Jo=E3o Cardoso wrote:

> But, as a first try to dyndrivers, I have no objections. However,
> plcore.c has to be changed in several places, plLibOpen() and
> plInitDispatchTable() at least.
>
> Joao

Thanks to you and Maurice for your quick responses.  I have gone ahead and
committed the appropriate changes (thanks, Joao, for that tip about plcore.=
c
changes), and the new $prefix/lib/plplot5.1.0/drivers install location seem=
s
to work well.  By all means give it a quick spin (after being sure to
completely clean out your old install location).

Alan

Re: [Plplot-devel] a few glitches in the test process. Also, a lot of test progress

[Maurice L. <mj...@ga...>]

Alan W. Irwin writes:
 > Geoffrey, Maurice, and Joao: you have all done some cross-platform testing
 > recently.  In the release notes can I describe that testing as Solaris/cc,
 > OSF1/Dec Alpha, and SuSe 7.1?  Please let me know if I have the qualifiers
 > right and/or whether I need more qualifiers to describe those platforms you
 > tested on.

I tried last week on the following systems with no problems:

zion$ uname -a
OSF1 zion.naka.jaeri.go.jp V4.0 564 alpha

naka-spf$ uname -a
IRIX64 naka-spf 6.5 04131233 IP35 mips

I'll give the latest another try soon to make sure nothing odd's come up.

-- 
Maurice LeBrun    mj...@ga...

Re: [Plplot-devel] Install location for the drivers directory?

[Maurice L. <mj...@ga...>]

Alan W. Irwin writes:
 > I would appreciate quick replies on this one since the question is holding
 > up the show.
 > 
 > I just noticed that the drivers are installed into $prefix/lib/drivers

Yikes.

 > Any objections if I change that to $prefix/lib/plplot5.1.0/drivers?

Please do.

-- 
Maurice LeBrun    mj...@ga...

Re: [Plplot-devel] Install location for the drivers directory?

[<jca...@in...>]

On Tuesday 29 January 2002 02:42, Alan W. Irwin wrote:
| I would appreciate quick replies on this one since the question is
| holding up the show.
|
| I just noticed that the drivers are installed into
| $prefix/lib/drivers
|
| Any objections if I change that to $prefix/lib/plplot5.1.0/drivers?

Drivers should be usable by several plplot releases, I think, so=20
tying then to a release-specific directory will disable this.

But, as a first try to dyndrivers, I have no objections. However,=20
plcore.c has to be changed in several places, plLibOpen() and=20
plInitDispatchTable() at least.

Joao

|
| Recall that we already install some of our files in
| $prefix/lib/plplot5.1.0/.
|
| I am concerned about name clashes if people are installing with
| prefix=3D/usr since other packages might use /usr/lib/drivers for
| something else.
|
|
| 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
| __________________________
|
|
| _______________________________________________
| Plplot-devel mailing list
| Plp...@li...
| https://lists.sourceforge.net/lists/listinfo/plplot-devel

Re: [Plplot-devel] a few glitches in the test process. Also, a lot of test progress

[<jca...@in...>]

On Monday 28 January 2002 21:47, Geoffrey Furnish wrote:
| Alan W. Irwin writes:
|  > Geoffrey, Maurice, and Joao: you have all done some
|  > cross-platform testing recently.  In the release notes can I
|  > describe that testing as Solaris/cc, OSF1/Dec Alpha, and SuSe
|  > 7.1?  Please let me know if I have the qualifiers right and/or
|  > whether I need more qualifiers to describe those platforms you
|  > tested on.
|
| I have built a clean checkout today on both SuSE-7.2 and Solaris
| 5.7.

Also on suse-7.0 and suse-7.2 (both with with static and dyndrivers)=20
and on SCO-3.2v5.0.4 (with-gcc, gmake, no shared libs, no dyndrivers).

I'm sorry to report that on a "OSF1 V4.0 alpha" I was not able to=20
complete the build, although I tried several  configuration options=20
(with/without gcc, with gmake, without-shlib,...):

=2E..
gcc   plrender.o -L. -lplplot  -o plrender  -lX11 -lm=20
/bin/ld:
Unresolved:
plD_dispatch_init_xw
plD_dispatch_init_xterm
=2E..
[all drivers]

Joao

|
| _______________________________________________
| Plplot-devel mailing list
| Plp...@li...
| https://lists.sourceforge.net/lists/listinfo/plplot-devel

[Plplot-devel] Install location for the drivers directory?

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

I would appreciate quick replies on this one since the question is holding
up the show.

I just noticed that the drivers are installed into $prefix/lib/drivers

Any objections if I change that to $prefix/lib/plplot5.1.0/drivers?

Recall that we already install some of our files in $prefix/lib/plplot5.1.0/.

I am concerned about name clashes if people are installing with prefix=/usr
since other packages might use /usr/lib/drivers for something else.


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] a few glitches in the test process. Also, a lot of test progress

[Geoffrey F. <fu...@ga...>]

Alan W. Irwin writes:
 > Geoffrey, Maurice, and Joao: you have all done some cross-platform testing
 > recently.  In the release notes can I describe that testing as Solaris/cc,
 > OSF1/Dec Alpha, and SuSe 7.1?  Please let me know if I have the qualifiers
 > right and/or whether I need more qualifiers to describe those platforms you
 > tested on.

I have built a clean checkout today on both SuSE-7.2 and Solaris 5.7.

[Plplot-devel] a few glitches in the test process. Also, a lot of test progress

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

OOPS.  In the commit message below I meant this change will be in 5.1.0 (not
5.0.1).

Also, I just discovered I screwed up the tag this morning (don't ask....
suffice it to say it was early in the morning) so actually the tag
v5_1_0_rc1 has only been applied  now (for the local version as of 8 AM this
morning plus this extra commited change.)

Aside from these minor glitches, the test progress has been great.  All
interactive tests are done on my home computer, and most of the "tmp"
non-interactive tests are done as well (psc, png, jpeg, and cgm done, with
plmeta to go).  After that I will do the "installed" tests on my home
machine (which will be much quicker because I can use diff rather than
visual inspection of the plots), then finish with build and "installed"
tests on the RH 7.1 and Solaris/Gnu platforms that are accessible to me.

Geoffrey, Maurice, and Joao: you have all done some cross-platform testing
recently.  In the release notes can I describe that testing as Solaris/cc,
OSF1/Dec Alpha, and SuSe 7.1?  Please let me know if I have the qualifiers
right and/or whether I need more qualifiers to describe those platforms you
tested on.

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
__________________________

On Mon, 28 Jan 2002, Alan W. Irwin wrote:

> Update of /cvsroot/plplot/plplot/cf
> In directory usw-pr-cvs1:/tmp/cvs-serv10258
>
> Modified Files:
> 	plplot-test.sh.in
> Log Message:
> 5.0.1.  Turn off familying option for cgm device since
> it is not necessary.
>
>
> _______________________________________________
> Plplot-cvs mailing list
> Plp...@li...
> https://lists.sourceforge.net/lists/listinfo/plplot-cvs
>

[Plplot-devel] bug in tk driver

[Geoffrey F. <fu...@ga...>]

Okay, I think I've reduced the problem to something I can confidently
blame on PLplot.

src/cplace> cat x.c
#include "plplot/plplot.h"

int main( int argc, char *argv[] )
{
    PLBufferingCB bcb;

    plParseOpts( &argc, argv, PL_PARSE_SKIP );
    plinit();

    bcb.cmd = PLESC_DOUBLEBUFFERING_ENABLE;
    pl_cmd( PLESC_DOUBLEBUFFERING, &bcb );

    pladv(0);
    plflush();
    pladv(0);

    plend();
}
src/cplace> gcc -I/lss/sw/fastflow/SuSE-7.2/include -o x x.c `plplot-config --libs`
src/cplace> ./x -dev tk
TCL command "plclient_link_end" failed:
         no application named "x"
src/cplace> 

If you comment out the line that sets the double buffering, the
plserver death goes away.

Since the bug is triggered by an obscure escape command, I think we
should not be deterred from releasing 5.1.0.  This needs to be fixed,
but is not something the average PLplot user is likely to encounter
early in thier PLplot experience.

-- 
Geoffrey Furnish         fu...@ga...

RE: [Plplot-devel] Deadline for 5.1.0 is Monday 8 AM PST = 16:00 UT

[Geoffrey F. <fu...@ga...>]

Geoffrey Furnish writes:
 >  > No idea.. I've never seen this.  BTW it may be informative to run including
 >  > the -debug option.  Then again it may not help.
 > 
 > I've demonstrated that the behavior is the same with snapshots from
 > 1/1/02 and from 9/1/01.  Therefore, I think it is clear that it is at
 > least not a recently introduced bug in PLplot.  Whether or not it is a
 > bug in PLplot at all, is yet to be determined, it may still be in my
 > code.  However, I haven't been able to find the cause yet.  Somehow it
 > seems that when running the tk driver, somehow my code cores and the
 > message is coming from plserver, saying the host app is dead.  When I
 > run the host code in the debugger, it sometimes claims to have ended
 > with a SIGPIPE.  Doesn't make any sense to me.  yet.  Using any other
 > device seems to work fine.

No, that's wrong.  It's plserver that's dieing, leaving the host app
up, but unable to communicate with it.  That explains why the host app
gets a sigpipe.  So now that makes sense.  But I don't understand
what's killing plserver.

Is there an easy way to run an app against a seperately started
plserver?  I seem to remember traffic on this list about this, but I
wasn't paying enough attention then to remember how to do it now.

-- 
Geoffrey Furnish         fu...@ga...

RE: [Plplot-devel] Deadline for 5.1.0 is Monday 8 AM PST = 16:00 UT

[Geoffrey F. <fu...@ga...>]

Maurice LeBrun writes:
 >  > I'm afraid I am writing to report some possibly discouraging news.
 >  > The Tk driver is not working for me right now.  I'm getting:
 >  > 
 >  > Packet send failed:
 >  >          pl_PacketSend -- error writing to fifo: bad file number
 >  > TCL command "plclient_link_end" failed:
 >  >          no application named "cplace_x86kaig"
 >  > Program aborted
 >  > 
 >  > when I run my code.  The tk driver starts, the plframe is displayed,
 >  > then it all vanishes in a puff of smoke, and 50% of the time or so, it
 >  > prints the above message, the other times it all simply vanishes
 >  > without a trace.
 > 
 > No idea.. I've never seen this.  BTW it may be informative to run including
 > the -debug option.  Then again it may not help.

I've demonstrated that the behavior is the same with snapshots from
1/1/02 and from 9/1/01.  Therefore, I think it is clear that it is at
least not a recently introduced bug in PLplot.  Whether or not it is a
bug in PLplot at all, is yet to be determined, it may still be in my
code.  However, I haven't been able to find the cause yet.  Somehow it
seems that when running the tk driver, somehow my code cores and the
message is coming from plserver, saying the host app is dead.  When I
run the host code in the debugger, it sometimes claims to have ended
with a SIGPIPE.  Doesn't make any sense to me.  yet.  Using any other
device seems to work fine.

-- 
Geoffrey Furnish         fu...@ga...

[Plplot-devel] release candidate 1 has been tagged with v5_1_0_rc1

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

Subject says it all....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
__________________________

[Plplot-devel] Re: Plplot on Mac OS X

[Don S. <sp...@or...>]

I noticed on the last Plplot-devel digest there was a question about 
use of plplot under Mac OS X.  I have recently converted my Macs over 
to OS X and was able to get Plplot up and running on them without too 
much difficulty.  This was much easier than under Mac OS 8 and 9 
where I was never able to get Plplot working in a reliable way.  The 
steps that I went through were:
- install the OS X developer tools.  This comes on a separate CD with 
the OS X operating system and includes such things as the gcc 
compiler, make, etc.
- install X-windows software (e.g., XDarwin, XFree86, etc.)
- I then opened up a terminal window and did the usual configure, 
make, and make install commands to build Plplot.  The only thing I 
remember having to do out of the ordinary here was to use the 
--prefix= flag to tell it where I wanted to install Plplot.  I was 
then able to go to the /tmp directory under Plplot and compile the 
examples.  They all seemed to work, using the X-Window display 
option.  At the time I did the installation, I didn't have png, jpeg 
or Tcl/Tk installed on my machine so those output options weren't 
available.  I now have at least the png and jpeg libraries installed 
so should be able to go back and rebuild Plplot to include those.
One approach that you might want to consider for making Plplot 
available to Mac OS X users is through the Fink Project 
(http://fink.sourceforge.net/) which I've recently started using. 
This project's purpose is to put together coherent distributions of 
UNIX software that OS X users can easily download and install using 
the fink package installer to manage the process.  (i.e., fink would 
know that Plplot needed XFree86, png, jpeg, Tcl/Tk, etc. and go out 
an get them as part of the installation process if they weren't 
already there).
-- 
_________________________________________________________
Donald A. Spong, Fusion Energy Theory, ORNL
Snail-mail: P. O. Box 2009
Oak Ridge, Tennessee  37831-8071
Phone: (865) 574-1304   FAX: (865) 576-7926
E-mail: sp...@or...
web page: http://www.ornl.gov/fed/Theory/stci/stellarator_theory.html
_________________________________________________________

RE: [Plplot-devel] Deadline for 5.1.0 is Monday 8 AM PST = 16:00 UT

[Geoffrey F. <fu...@ga...>]

Alan W. Irwin writes:
 > On Sat, 26 Jan 2002, Geoffrey Furnish wrote:
 > > [...]Now, you might suspect its just my code.  And that could well be.
 > > However, it runs to completion with no upset using many other devices,
 > > including xwin, ps, and plmeta.  With a metafile, I can render to
 > > device tk.  Also the plplot C examples work with dev tk.  So I realize
 > > this is a pretty inconclusive set of data.
 > 
 > The C examples with tk driver may not be adequately exercising the code.
 > Therefore, I think you should exercise every Tk example to see if there are
 > any obvious problems on your system.  Here is my script for doing this from
 > plplot/tmp:

Thanks.  The script was helpful, things seem to be working in the core
distribution.  But I agree with your suspicion that my particular
client code is probably doing things wityh the API that the examples
don't.  However, I haven't been able to pin it down yet to a single
specific problem, and realistically, do not expect to have such
closure in the next 14 hours.  So, unless someone else announces
similar problems, I think you should just go ahead.

I did manage to build my code with a completely different compiler,
just to try to eliminate the "bad code gen" prospect.  The same
failure occurs even when the client code is built with the latst gcc,
3.0.3.  So, the plot thickens.

Realistically, I think its going to take a timescale of days for me to
get to the bottom of this, so you should just proceed.  The problem
may well turn out to have nothing to do with PLplot.

More when I know more.

-- 
Geoffrey Furnish         fu...@ga...

RE: [Plplot-devel] Deadline for 5.1.0 is Monday 8 AM PST = 16:00 UT

[Maurice L. <mj...@ga...>]

Geoffrey Furnish writes:
 > Alan W. Irwin writes:
 >  > Hello everybody.  Since our efforts to release 5.1.0 seem to have stalled,
 >  > I'm reinitiating the test/release plan.  
 > 
 > Thanks Alan for restarting the count down.

Sorry my efforts have slowed, but I may yet get some more stuff sorted out in
time for inclusion for tomorrow's release.

 > I'm afraid I am writing to report some possibly discouraging news.
 > The Tk driver is not working for me right now.  I'm getting:
 > 
 > Packet send failed:
 >          pl_PacketSend -- error writing to fifo: bad file number
 > TCL command "plclient_link_end" failed:
 >          no application named "cplace_x86kaig"
 > Program aborted
 > 
 > when I run my code.  The tk driver starts, the plframe is displayed,
 > then it all vanishes in a puff of smoke, and 50% of the time or so, it
 > prints the above message, the other times it all simply vanishes
 > without a trace.

No idea.. I've never seen this.  BTW it may be informative to run including
the -debug option.  Then again it may not help.

-- 
Maurice LeBrun    mj...@ga...

[Plplot-devel] "driver" misnomer changed to "device" in plplot-test.sh and friends

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

That change is important for the tests starting tomorrow.

Invocation of plplot-test.sh changed from, e.g.,

./plplot-test.sh --driver=png

to

./plplot-test.sh --device=png

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
__________________________