plplot-devel — Technical discussion about PLplot development

[Plplot-devel] Have Geoffrey's fidelity concerns been addressed?

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

Geoffrey, I haven't yet had a chance to test Maurice's recent character
precision improvements to see whether they improve the cross-platform
reproducibility of the PLplot examples.  However, I am hoping for a
substantial improvement.  Also, I am wondering if his changes have had an
impact on the fidelity problems you were having before with illegible
characters (IIRC) in a zoomed environment?

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] Can we get the plframe widget working with python/Tkinter now?

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

The only limitation that I am aware of for the normal python build is we are
not yet able to use the plframe widget with python/Tkinter.  I would like
to see this limitation removed.

Geoffrey, this note is a reminder that our tkdemos.tcl and tcldemos.tcl
examples now work under tclsh and wish using dynamic loading.  In the past
you have said that was a prerequisite for getting plframe to work with
python/Tkinter.

I just tried to get the old set of python scripts (pytkdemo, Plframe.py, and
the old x??.py examples that are called by pytkdemo) to work.  Obviously
bit-rot has set in, but by doing some obvious hacks (pl --> plplot, etc.) I
got this far using a test version of pytkdemo called test.py:

./test.py
Building the Python Tk demo application main window.
Traceback (most recent call last):
  File "./test.py", line 174, in ?
    demo = DemoApp()
  File "./test.py", line 18, in __init__
    s.create_widgets()
  File "./test.py", line 27, in create_widgets
    s.plf = PlXframe(s, width="17c", height="17c" )
  File "./Plframe.py", line 127, in __init__
    s.plf = Plframe( s, kw )
  File "./Plframe.py", line 37, in __init__
    Widget.__init__( self, master, 'plframe', cnf, kw )
  File "/usr/lib/python2.1/lib-tk/Tkinter.py", line 1756, in __init__
    self.tk.call(
TclError: invalid command name "plframe"

So what is the next step (perhaps some additional special wrapper functions
need to be written for plplot_widgetmodule.c?) to get the plframe widget
working in an environment where anything from the plplplot libraries can be
dynamically loaded into python/Tkinter?

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] Prescriptions for running tkdemos.tcl under wish and tcldemos.tcl under tclsh

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

On Fri, 27 Sep 2002, Maurice LeBrun wrote:

> [...]Now it always checks the known install location first.

Thanks, Maurice.  Everything works fine here now without having to set the
PATH or PL_LIBRARY environment variables.

I have put the new prescriptions into README.tcldemos and README.tkdemos.

Alan

Re: [Plplot-devel] Prescriptions for running tkdemos.tcl under wish and tcldemos.tcl under tclsh

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

Alan W. Irwin writes:
 > > I don't know what you mean here.  Could you write out the steps that generate
 > > the error so I can look at it?
 > 
 > (1) Clean build and install from CVS (same steps I detailed earlier).
 > ...

OK, thanks for the prescription.  I wasn't seeing the problem b/c my wish runs
out of the same $prefix as plplot, and the existing initialization logic was
treating the installed plplot image as basically a tcl extension that lived in
the same tree, except when building ("." is checked last) or if PL_LIBRARY is
set.  Now it always checks the known install location first.

-- 
Maurice LeBrun    mj...@ga...
Research Organization for Information Science and Technology of Japan (RIST)

Re: [Plplot-devel] Prescriptions for running tkdemos.tcl under wish and tcldemos.tcl under tclsh

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

On Thu, 26 Sep 2002, Maurice LeBrun wrote:

> Alan W. Irwin writes:

>  > (2) As reported yesterday, I could not get tcldemos.tcl to work under tclsh
>  > without setting PL_LIBRARY to /usr/local/plplot/lib/plplot5.1.0/tcl.  I
>  > presume this is also a requirement for running tkdemos.tcl under wish
>  > and runAllDemos.tcl under wish.  But should it be?  Under tclsh and
>  > wish I already execute
>  >
>  > lappend auto_path /usr/local/plplot/lib/plplot5.1.0
>  >
>  > Thus, shouldn't we be able to find the location of everything internally by
>  > scanning through the directories in auto_path?  We have been through so many
>  > variations of the tcl/tk install that I cannot remember clearly any more,
>  > but I believe there were some variations on the theme that did not require
>  > setting PL_LIBRARY, and I hope we can get back to that.
>
> I don't know what you mean here.  Could you write out the steps that generate
> the error so I can look at it?

(1) Clean build and install from CVS (same steps I detailed earlier).

(2) cp -a /usr/local/plplot/lib/plplot5.1.0/examples/ /tmp

(3) cd /tmp/example/tk

(4) Put /usr/local/plplot/bin on PATH (required for -dev tk.  not sure it is
relevant in this case, but that is what I did)

(5) Make sure PL_LIBRARY *not* set (unsetenv PL_LIBRARY under tcsh)

(6) package require Pltk craps out under wish.

wish
% lappend auto_path /usr/local/plplot/lib/plplot5.1.0
/usr/lib/tcl8.3 /usr/lib /usr/lib/tk8.3 /usr/local/plplot/lib/plplot5.1.0
% package require Pltk
Can't find a usable plplot.tcl in the following directories:
    /usr/lib/plplot5.1.0/tcl /usr/lib/plplot5.1.0/tcl /lib/plplot5.1.0/tcl
/usr/library /library /plplot5.1.0/tcl/library /plplot5.1.0/tcl/library



This probably means that plplot wasn't installed properly.

..................

OTOH if I set PL_LIBRARY first, everything works fine.

setenv PL_LIBRARY /usr/local/plplot/lib/plplot5.1.0/tcl
wish
% lappend auto_path /usr/local/plplot/lib/plplot5.1.0
/usr/lib/tcl8.3 /usr/lib /usr/lib/tk8.3 /usr/local/plplot/lib/plplot5.1.0
% package require Pltk
5.1.0
% source tkdemos.tcl
% 1
% 2

etc.

The same story occurs (PL_LIBRARY needs to be set) with tcldemos.tcl under
tclsh (and presumably runAllDemos.tcl under wish).

Do you get good results without PL_LIBRARY being set?  If so, perhaps there
is a bug in my tcl/tk version (which is 8.3.3).

Alan

Re: [Plplot-devel] Prescriptions for running tkdemos.tcl under wish and tcldemos.tcl under tclsh

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

Alan W. Irwin writes:
 > Here are some current limitations of the tcl/tk PLplot install I would like
 > to see removed in the long term.
 > 
 > (1) Must have /usr/local/plplot/bin in PATH in order for -dev tk to work. tk
 > is the only device with this requirement.  As I understand it, this
 > requirement is needed because the tk device starts its own instance of
 > plserver, and it needs to be able to find it.  Doesn't the PLplot library
 > already know where everything is (for both the plplot/tmp location and
 > installed location) so it could find plserver without having to rely on the
 > PATH?  Obviously you need the PATH set if you are doing lots of plrendering
 > or plserving.  But there are many cases of using PLplot where you just
 > simply want to use -dev tk with, e.g, the python examples or C examples so
 > it would be nice to remove this limitation of requiring the PATH to be set
 > for -dev tk.

I agree, the tk driver should consider other places than just the existing
path.  In fact I now think the path should be checked last (if at all), with
the search order going as:
 - the current dir, *if* it looks like a plplot build (tmp) dir
   (not sure how to check that but it should be possible)
 - the install dir
 - existing path

Will put it on my to-do list, shouldn't be that much work (famous last words)
but might require some thought.

 > (2) As reported yesterday, I could not get tcldemos.tcl to work under tclsh
 > without setting PL_LIBRARY to /usr/local/plplot/lib/plplot5.1.0/tcl.  I
 > presume this is also a requirement for running tkdemos.tcl under wish
 > and runAllDemos.tcl under wish.  But should it be?  Under tclsh and
 > wish I already execute
 > 
 > lappend auto_path /usr/local/plplot/lib/plplot5.1.0
 > 
 > Thus, shouldn't we be able to find the location of everything internally by
 > scanning through the directories in auto_path?  We have been through so many
 > variations of the tcl/tk install that I cannot remember clearly any more,
 > but I believe there were some variations on the theme that did not require
 > setting PL_LIBRARY, and I hope we can get back to that.

I don't know what you mean here.  Could you write out the steps that generate
the error so I can look at it?

-- 
Maurice LeBrun    mj...@ga...
Research Organization for Information Science and Technology of Japan (RIST)

Re: [Plplot-devel] Prescriptions for running tkdemos.tcl under wish and tcldemos.tcl under tclsh

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

On Wed, 25 Sep 2002, Maurice LeBrun wrote:

> Try again now, with fixes under cf/.

OK.  Used clean start (rm -rf /usr/local/plplot/*; make clean + inspection
of plplot/tmp to make sure it really worked; cvs update; make configure;
configure; make; make install).  Copied the installed examples to
/tmp/examples (note this is an absolute location and nothing to do with
plplot/tmp).

With some limitations to be discussed below -dev tk worked fine as well as
tcldemos.tcl, tkdemos.tcl, and runAllDemos.tcl under the appropriate recipes
published yesterday. Thanks, Maurice, for going "above and beyond" on this
one during a time when you expected to be free of PLplot.

Here are some current limitations of the tcl/tk PLplot install I would like
to see removed in the long term.

(1) Must have /usr/local/plplot/bin in PATH in order for -dev tk to work. tk
is the only device with this requirement.  As I understand it, this
requirement is needed because the tk device starts its own instance of
plserver, and it needs to be able to find it.  Doesn't the PLplot library
already know where everything is (for both the plplot/tmp location and
installed location) so it could find plserver without having to rely on the
PATH?  Obviously you need the PATH set if you are doing lots of plrendering
or plserving.  But there are many cases of using PLplot where you just
simply want to use -dev tk with, e.g, the python examples or C examples so
it would be nice to remove this limitation of requiring the PATH to be set
for -dev tk.

(2) As reported yesterday, I could not get tcldemos.tcl to work under tclsh
without setting PL_LIBRARY to /usr/local/plplot/lib/plplot5.1.0/tcl.  I
presume this is also a requirement for running tkdemos.tcl under wish
and runAllDemos.tcl under wish.  But should it be?  Under tclsh and
wish I already execute

lappend auto_path /usr/local/plplot/lib/plplot5.1.0

Thus, shouldn't we be able to find the location of everything internally by
scanning through the directories in auto_path?  We have been through so many
variations of the tcl/tk install that I cannot remember clearly any more,
but I believe there were some variations on the theme that did not require
setting PL_LIBRARY, and I hope we can get back to that.

Alan

Re: [Plplot-devel] Prescriptions for running tkdemos.tcl under wish and tcldemos.tcl under tclsh

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

Maurice LeBrun writes:
 > Maurice LeBrun writes:
 >  > Alan W. Irwin writes:
 >  >  [various problems]
 >  > 
 >  > I just tried this and after several iterations of reduction, finally
 >  > reproduced the problem on a fresh install.  It goes away for me after
 >  > I type "make install" the second time.  So apparently some problem relinking
 >  > the dynamic drivers against the install libs.  I'll check to see if it was
 >  > something I did in my last changes.
 > 
 > Found it!  My bad.  Will fix. :)

Try again now, with fixes under cf/.

-- 
Maurice LeBrun    mj...@ga...
Research Organization for Information Science and Technology of Japan (RIST)

Re: [Plplot-devel] Prescriptions for running tkdemos.tcl under wish and tcldemos.tcl under tclsh

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

Maurice LeBrun writes:
 > Alan W. Irwin writes:
 >  [various problems]
 > 
 > I just tried this and after several iterations of reduction, finally
 > reproduced the problem on a fresh install.  It goes away for me after
 > I type "make install" the second time.  So apparently some problem relinking
 > the dynamic drivers against the install libs.  I'll check to see if it was
 > something I did in my last changes.

Found it!  My bad.  Will fix. :)

-- 
Maurice LeBrun    mj...@ga...
Research Organization for Information Science and Technology of Japan (RIST)

Re: [Plplot-devel] Prescriptions for running tkdemos.tcl under wish and tcldemos.tcl under tclsh

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

Alan W. Irwin writes:
 [various problems]

I just tried this and after several iterations of reduction, finally
reproduced the problem on a fresh install.  It goes away for me after
I type "make install" the second time.  So apparently some problem relinking
the dynamic drivers against the install libs.  I'll check to see if it was
something I did in my last changes.

 > I don't care that strongly if we install our tcl/tk-related stuff in
 > $prefix/lib/tcl8.3/plplot5.1.0 or $prefix/lib/plplot5.1.0 or
 > $prefix/lib/pltcl5.1.0, but I think we should make the choice consistently
 > for all our tcl/tk related files and thus get rid of the tcl subdirectory in
 > the install location. If we also drop the tcl subdirectory that was hacked
 > into tclAPI.c, I suspect our current tcl/tk PLplot install troubles might be
 > reduced or eliminated.

The $prefix/lib/plplot5.1.0/tcl dir must be in auto_path, as that's where all
the tcl support routines are.  When loading dynamically, the "package require
xxx" does this.  When using our custom tclsh/wish, it's done internally.  It
can't be dropped.

-- 
Maurice LeBrun    mj...@ga...
Research Organization for Information Science and Technology of Japan (RIST)

Re: [Plplot-devel] Prescriptions for running tkdemos.tcl under wish and tcldemos.tcl under tclsh

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

On Wed, 25 Sep 2002, Alan W. Irwin wrote:

> I mention the current directory only to tell you that trying to
> execute the examples from the installed examples location does not work.

OOPS.  That came out wrong.  What I meant to say is I got the same symptoms
(mostly good, but with the few problems I mentioned) whether executing from
the installed examples location or from a copied version in /tmp/examples.

Alan

Re: [Plplot-devel] Prescriptions for running tkdemos.tcl under wish and tcldemos.tcl under tclsh

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

Thanks, Maurice, for working on this when you expected to be free of Plplot
for a while.  (Actually, I hoped to be free of PLplot myself for the last 6
weeks or so, but curiosity about what is happening keeps bringing me
back....;-))

Unfortunately, there are still some tcl/tk problems on my system for the
installed version of PLplot.  I believe this is something that does not
necessarily need to be addressed right away, but let me make my comments now
while my recent experience is fresh in my mind.

I removed the installed version (prefix of /usr/local/plplot) and then
reinstalled to that same location using the latest CVS version of plplot.

On my system tcl/tk is not smart enough to pick up the prefix
(/usr/local/plplot in my case) from either auto_path or from the current
directory.  (I mention the current directory only to tell you that trying to
execute the examples from the installed examples location does not work.  In
any case, I want an installed system where I can copy the examples anywhere
and execute them with tcl/tk able to find everything it needs.)

To get around this limitation I set the environment variable PL_LIBRARY to
/usr/local/plplot/lib/plplot5.1.0.  Then the tclsh recipe below works
fine with tcl examples copied to any directory.  However, the wish recipe
for tkdemos.tcl pooped out with the following error message:
wish
% lappend auto_path /usr/local/plplplot/lib/plplot5.1.0
% package require Pltk
5.1.0
% source tkdemos.tcl
invalid command name "plstdwin"

During the course of my testing I also found a different problem with the
installed version.  (In this case I had copied the installed examples to
/tmp, but the same problem occurred if I used the installed examples in
situ.)

cd /tmp/examples/c
make x01c
./x01c -dev tk
TCL command "plclient_init" failed:
         invalid command name "plclient_init"
TCL command "plclient_link_end" failed:
         invalid command name "plclient_link_end"

Not that long ago the tk device worked fine for installed Plplot, but I
don't know what recent change is the culprit.

We have been fighting this "installed tcl/tk" problem for months. At my
instigation Vince hacked PLPLOT_EXTENDED_SEARCH into tclAPI.c to try and
beat the system to use a subdirectory tcl, but in retrospect I think this
was a mistake.  Instead, I think we should just install all the tcl scripts
right in a single directory where tcl/tk normally expects to find them.

I checked on my Debian system for the location of pkgIndex.tcl files.

locate pkgIndex.tcl |grep '^/usr/lib'
/usr/lib/blt2.4/pkgIndex.tcl
/usr/lib/expect5.32/pkgIndex.tcl
/usr/lib/itcl3.1/pkgIndex.tcl
/usr/lib/itk3.1/pkgIndex.tcl
/usr/lib/iwidgets3.1.0/pkgIndex.tcl
/usr/lib/tcl8.3/http1.0/pkgIndex.tcl
/usr/lib/tcl8.3/http2.3/pkgIndex.tcl
/usr/lib/tcl8.3/msgcat1.1/pkgIndex.tcl
/usr/lib/tcl8.3/opt0.4/pkgIndex.tcl
/usr/lib/tcl8.3/tcltest1.0/pkgIndex.tcl
/usr/lib/tk8.3/pkgIndex.tcl

So at first glance it seems there is a choice between making the install
location a subdirectory of /usr/lib or of /usr/lib/tcl8.3.  However, a
further check indicates the latter directories are all part of tcl8.3 and
not "foreign" packages so I concur with Maurice's recent decision to install
our pkgIndex.tcl in $prefix/lib/plplot5.1.0. But I suggest we might want to
go further and create a new directory called $prefix/lib/pltcl5.1.0 and put
our pkgIndex.tcl and everything currently in $prefix/lib/plplot5.1.0/tcl in
there just to reduce the unaesthetic clutter in $prefix/lib/pltcl5.1.0 of
a mixture of tcl files and non-tcl subdirectories.

I don't care that strongly if we install our tcl/tk-related stuff in
$prefix/lib/tcl8.3/plplot5.1.0 or $prefix/lib/plplot5.1.0 or
$prefix/lib/pltcl5.1.0, but I think we should make the choice consistently
for all our tcl/tk related files and thus get rid of the tcl subdirectory in
the install location. If we also drop the tcl subdirectory that was hacked
into tclAPI.c, I suspect our current tcl/tk PLplot install troubles might be
reduced or eliminated.

Maurice is presently out of time and so am I so if somebody with time on his
hands (heh) wants to take over now and fix this tcl/tk PLplot install
problem for Linux with the install location of his choice, that would be
great.

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 Wed, 25 Sep 2002, Maurice LeBrun wrote:

> The changes I've just committed enable easy loading of all three kinds, from
> both the tmp/ dir and the install tree.  I've given the prescription for the
> tmp/ dir only; in the install tree you must:
>  - change the auto_path setting appropriately (or use a tclsh/wish
>    under the same $prefix)
>  - first cd to the appropriate examples/xxx dir.
>
> Under plplot/tmp/:
>
> tclsh
> lappend auto_path .
> package require Pltcl
> plinit
> source tcldemos.tcl
> 1
> 2
> ...
>
> wish
> lappend auto_path .
> package require Pltk
> source tkdemos.tcl
> 1
> 2
> ...
>
> wish
> lappend auto_path .
> package require Plplotter
> source runAllDemos.tcl
> <control from gui>

Re: [Plplot-devel] Prescriptions for running tkdemos.tcl under wish and tcldemos.tcl under tclsh

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

The changes I've just committed enable easy loading of all three kinds, from
both the tmp/ dir and the install tree.  I've given the prescription for the
tmp/ dir only; in the install tree you must:
 - change the auto_path setting appropriately (or use a tclsh/wish
   under the same $prefix)
 - first cd to the appropriate examples/xxx dir.

Under plplot/tmp/:

tclsh
lappend auto_path .
package require Pltcl
plinit
source tcldemos.tcl
1
2
...

wish
lappend auto_path .
package require Pltk
source tkdemos.tcl
1
2
...

wish
lappend auto_path .
package require Plplotter
source runAllDemos.tcl
<control from gui>

The way this is done is as follows.

Since there are 3 ways of loading, there are 3 pkgIndex.tcl.in files, one for
each dir under bindings/.  They are very similar but have slightly different
parameters, and should be easy enough to understand & modify.

I gave them the ".in" extension to distinguish them from ordinary tcl files --
I don't want them in the tclIndex or being installed with the others.  Rather,
they are concatenated to get the real pkgIndex.tcl file.  So this is done
during configure to get the pkgIndex.tcl under tmp/, and again during install
to get the one under $prefix/lib/plplot$version/.

-- 
Maurice LeBrun    mj...@ga...
Research Organization for Information Science and Technology of Japan (RIST)

Re: [Plplot-devel] Prescriptions for running tkdemos.tcl under wish and tcldemos.tcl under tclsh

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

Alan W. Irwin writes:
 > I confirm those display problems on my system with the above recipe.  I
 > think the problem is that you have set up everything so that "package
 > require Plplotter" indirectly executes
 > 
 > load ./drivers/tkwind_drv.so Plplotter
 > 
 > (see pkgIndex.tcl).  That is perfect for source "runAllDemos.tcl", but that
 > is not correct for either "source tkdemos.tcl" (under wish) or "source
 > tcldemos.tcl" (under tclsh).

Hmm.. ok that makes sense, although it is a bit odd.  One of these days I'll
have to get around to merging the two forks if possible.

 > Instead, for the tkdemos.tcl case
 > "package require Pltk" should indirectly execute
 > 
 > load ./drivers/tkd_drv.so Pltk
 > 
 > (very different from the load command above).
 > 
 > Similarly, for the tcldemos.tcl case
 > "package require Pltcl" should indirectly execute
 > 
 > load ./libplplottcltkd.so.5.1.0 Pltcl
 > 
 > Maurice, can you put that required extra functionality (with generalization
 > to the installed case once the plplot/tmp case works) into pkgIndex.tcl? (I
 > tried more or less blindly following what you did for the "package require
 > Plplotter" case, but ran into errors due to my extremely limited knowledge
 > of tcl).

So much for my break from plplot until next week.. :/

I was curious & thought it'd be a quick fix so I started work on it already.
I've devised a nice way to handle everything, still in the testing phase,
a bit more work than a "quick fix" but oh well, kind of cool at least.
Should get it committed tonight.

-- 
Maurice LeBrun    mj...@ga...
Research Organization for Information Science and Technology of Japan (RIST)

[Plplot-devel] plfreetype orientation troubles

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

Andrew, your newly checked in plfreetype.c still has some orientation
troubles for non-integral -ori values even when a unity aspect ratio is
forced.  I double-checked that the same troubles also occurred for the
plfreetype.c test version I was working with yesterday.  However, I don't
think it is worth worrying about the plfreetype.c orientation problems until
Maurice can find a fix (after October 1st) for the parallelogram problem in
the core.

I suspect what has been going on is that the handling of both the rotation
and placement of character strings for postscript (in ps.c) and freetype
fonts has some heuristic factors put in to try and compensate for the core
problem for integral -ori values, but of course such a band-aid solution is
unlikely to work for non-integral -ori values.  But once Maurice solves the
core problem that is causing the box to turn into a parallelogram, I think
the handling of string rotation and placement for non-Hershey fonts will
become much easier to understand and programme regardless of orientation.

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] Prescriptions for running tkdemos.tcl under wish and tcldemos.tcl under tclsh

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

On Tue, 24 Sep 2002, Maurice LeBrun wrote:

> The changes I just committed also allow the following style of loading,
> from plplot/tmp:
>
> wish
> lappend auto_path .
> package require Plplotter
> source tkdemos.tcl
> 1
> 2
> ...
>
> The only problem is that, on my system, the resulting window has some
> redisplay problems.

I confirm those display problems on my system with the above recipe.  I
think the problem is that you have set up everything so that "package
require Plplotter" indirectly executes

load ./drivers/tkwind_drv.so Plplotter

(see pkgIndex.tcl).  That is perfect for source "runAllDemos.tcl", but that
is not correct for either "source tkdemos.tcl" (under wish) or "source
tcldemos.tcl" (under tclsh).

Instead, for the tkdemos.tcl case
"package require Pltk" should indirectly execute

load ./drivers/tkd_drv.so Pltk

(very different from the load command above).

Similarly, for the tcldemos.tcl case
"package require Pltcl" should indirectly execute

load ./libplplottcltkd.so.5.1.0 Pltcl

Maurice, can you put that required extra functionality (with generalization
to the installed case once the plplot/tmp case works) into pkgIndex.tcl? (I
tried more or less blindly following what you did for the "package require
Plplotter" case, but ran into errors due to my extremely limited knowledge
of tcl).

Alan

Re: [Plplot-devel] Prescriptions for running tkdemos.tcl under wish and tcldemos.tcl under tclsh

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

Alan W. Irwin writes:
 > I just made an easy fix to the version problem (5.1.0 is now returned from
 > package require wish command rather than the incorrect 4.99 that was
 > returned previously).
 > 
 > I will repeat the prescriptions again from my last post with this subject to
 > remind the tcl/tk experts to try this.  Also, please note the final
 > paragraph.
 > 
 > Here is the wish/tkdemos.tcl prescription from plplot/tmp:
 > 
 > wish
 > 
 > load drivers/tkd_drv.so Pltk
 > package require Pltk
 > source tkdemos.tcl
 > 
 > 1
 > 2
 > 3
 > .....

Finally got around to working on this again.  Probably about my last
bit of plplot work for the next week or so.

The changes I just committed also allow the following style of loading,
from plplot/tmp:

wish
lappend auto_path .
package require Plplotter
source tkdemos.tcl
1
2
...

The only problem is that, on my system, the resulting window has some
redisplay problems -- sometimes I have to e.g. switch desktops then switch
back to have it display.  When I load according to Alan's prescription,
that doesn't happen.  Very odd!  Anyone have any ideas?

 > Also some changes will have to be done to the files which set up tcl/tk so
 > that the package require commands will work without the prior load commands
 > both for the plplot/tmp and installed versions.  However, I will leave that
 > internal tcl/tk configuration issue to the tcl/tk experts here.

The above prescription will now also work for the installed image.
cd to the $prefix/lib/plplot5.1.0/examples/tk/ dir, then type:

wish
lappend auto_path <prefix>/lib/plplot5.1.0
package require Plplotter 
source tkdemos.tcl 
1 
2 
... 

(note if your wish is also located under the same $prefix, you can omit 
the lappend to auto_path)

Note the install location of pkgIndex.tcl has changed -- it's now under
$prefix/lib/plplot5.1.0/ as is customary (see itcl, etc).  So delete the
old copy from your install tree.

I got the same redisplay problems running out of the install location.
Aside from this, I believe everything is working as it should be now.

-- 
Maurice LeBrun    mj...@ga...
Research Organization for Information Science and Technology of Japan (RIST)

Re: [Plplot-devel] Bug: non-integral -ori values turn box into parallelogram

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

Alan W. Irwin writes:
 > I have recently run across an orientation bug for non-integral -ori values.
 > If I try
 > 
 > ./x12c -dev xwin -ori 0.2
 > 
 > and hold a piece of paper up to the screen it is immediately obvious that
 > the box has turned into a parallelogram with corner angles which are
 > noticeably different from 90 deg.  [...]

I'll look into it when I get a chance.  This stuff is a bit on the hairy side
so I'm not sure there's an easy fix.  I was pretty happy to get integral
rotations working; the non-integral ones were just a freebie and something to
play with.  But now that I've "recovered" from the experience of setting it
all up, maybe I can come up with the 100% fix.  Not till after the 1st tho.

-- 
Maurice LeBrun    mj...@ga...
Research Organization for Information Science and Technology of Japan (RIST)

Re: [Plplot-devel] "x01c -dev xwin -locate" fails

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

Jo=E3o Cardoso writes:
 >=20
 > Hi,
 >=20
 > I noticed that plGetCursor() fails in the xwin driver, as it returns=
 as=20
 > soon as there is a mouse movement in the plot window, even without=20=

 > mouse clicking/draging.
 >=20
 > Is this reprodutible in your systems?

It's working as expected on my system (RH7.3).

--=20
Maurice LeBrun    mj...@ga...
Research Organization for Information Science and Technology of Japan (=
RIST)

[Plplot-devel] Re: [Plplot-general] X_CreateWindow

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

(taken off the main list for followups)

James Clark Calloway writes:
 > 
 > I have an alpha w/ OSF1 and I am getting
 > the following error when I try to do
 > plots:
 > 
 > X Error of failed request:  BadMatch (invalid parameter attributes)
 > Major opcode of failed request:  1	(X_CreateWindow)
 > Serial number of failed request:  12
 > Current serial number in output stream:  16
 > 
 > 
 > I tried the AC_DEFINE(USE_DEFAULT_VISUAL) mentioned 
 > in a previous letter but it did not resolve the problem.
 > Anyone have any idea ?

Please try the following change to xwin.c.  In InitMain() add
the declaration:

      Window root;
      XSizeHints hint;
      XSetWindowAttributes attr;           /* new */
      int x, y;
      U_INT width, height, border, depth;
      char header[80];

and change the last arg in the XCreateWindow call from:

                       hint.x, hint.y, hint.width, hint.height,
                       dev->border, xwd->depth,
                       InputOutput, xwd->visual,
                       0, NULL );
  

to:

                       hint.x, hint.y, hint.width, hint.height,
                       dev->border, xwd->depth,
                       InputOutput, xwd->visual,
                       0, &attr );

Let me know how it goes.

-- 
Maurice LeBrun    mj...@ga...
Research Organization for Information Science and Technology of Japan (RIST)

Re: [Plplot-devel] python works fine on RH 7.3

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

On Sun, 15 Sep 2002, I previously wrote about the possibility of changing
from swig-1.3.1[1-3] to later (e.g., 1.3.15) in the unstable series which will
lead to a stable 2.0 version of swig.

FYI, I have discussed this with Gary, and neither one of us has the time
currently to do the required adaptation to swig-1.3.15.  Thus, the decision
is to stick with swig-1.3.1[1-3] for now, and I (with possibly some help
from Gary) will attempt to do the required adaptation to swig-2.0 when it
comes out.

So the current status is Gary (RedHat, MacOS X, and windows), I (Debian and
RH 7.3), Maurice (RH 7.3), and Olof (windows) have built and used the new
python interface with no reported problems using swig-1.3.1[1-3].  In
contrast Geoffrey and Joao did have problems presumably because they were
using something different from swig-1.3.1[1-3].  Will you guys please try
again with a swig version in the required range (1.3.11 works for me), and
let me know if there are any problems?

Alan

[Plplot-devel] Prescriptions for running tkdemos.tcl under wish and tcldemos.tcl under tclsh

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

I just made an easy fix to the version problem (5.1.0 is now returned from
package require wish command rather than the incorrect 4.99 that was
returned previously).

I will repeat the prescriptions again from my last post with this subject to
remind the tcl/tk experts to try this.  Also, please note the final
paragraph.

Here is the wish/tkdemos.tcl prescription from plplot/tmp:

wish

load drivers/tkd_drv.so Pltk
package require Pltk
source tkdemos.tcl

1
2
3
.....

Here is the tclsh/tcldemos.tcl prescription from plplot/tmp:

tclsh

load ./libplplottcltkd.so.5.1.0 Pltcl
package require Pltcl
plinit
source tcldemos.tcl

1
2
3
.....

Also some changes will have to be done to the files which set up tcl/tk so
that the package require commands will work without the prior load commands
both for the plplot/tmp and installed versions.  However, I will leave that
internal tcl/tk configuration issue to the tcl/tk experts here.

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] "x01c -dev xwin -locate" fails

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

Hi,

I noticed that plGetCursor() fails in the xwin driver, as it returns as=20
soon as there is a mouse movement in the plot window, even without=20
mouse clicking/draging.

Is this reprodutible in your systems?

Joao

[Plplot-devel] Bug: non-integral -ori values turn box into parallelogram

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

I have recently run across an orientation bug for non-integral -ori values.
If I try

./x12c -dev xwin -ori 0.2

and hold a piece of paper up to the screen it is immediately obvious that
the box has turned into a parallelogram with corner angles which are
noticeably different from 90 deg.  This problem also occurs for the psc and
png devices.  If you would like an extreme case of this problem try a large
aspect ratio such as

./x12c -dev xwin -ori 0.2 -a 10

If you want the problem to completely disappear try an aspect ratio of unity.

./x12c -dev xwin -ori 0.2 -a 1

I suspect the core is doing something dumb like applying a rotation
transformation to normalized device coordinates (whose x,y ratio varies from
true device coordinates by typically a factor of 3/4, but you can make this
10 or unity by the above manipulations of the aspect ratio) rather than
actual device coordinates.  However, there is no way I will ever understand
the core rotation code so I cannot fix this myself.

Maurice, could you have a look at this to see if there is a quick fix
to this "parallelogram" bug?

BTW, the reason I ran into this was I was comparing how the postscript fonts
in the ps.c code and freetype.c (Andrew's version which he should shortly be
checking in) handled non-integral -ori values.  The results currently look
ugly because both codes rotate the fonts properly (x and y labels written at
90 deg with respect to each other) while the core code does not rotate the
plot properly (box angles not 90 deg unless the aspect ratio is unity).

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] precision problems in c_plmtex()

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

I am very happy you have moved to normalized device coordinates for
character placement and along the way found and fixed some major precision
problems.  I don't care if your changes break consistency with prior results
(such consistency was always getting broken in any case).  Precision
problems for character stroke placement have long been my number one suspect
for why I could not get an absolutely clean comparisons for all examples
from one architecture/compiler to the next or from an old release to a new
one or from one front end to another.

But now that you have found and fixed the problems in the character code
where the precision was getting clobbered, I am sure the consistency in
results will be much improved, and any remaining precision problems (if they
exist) will be much easier to find.

Thanks for this big breakthrough, Maurice!

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
__________________________