plplot-general — PLplot user discussion group

Re: [Plplot-general] Question on PlcGrid2

[Jerry <lan...@qw...>]

On Dec 6, 2006, at 3:32 AM, Jerry wrote:

> plot.h contains the following. Is this correct? Is there one extra
> array or one too few dimension sizes or do I need to study the docs
> more?
> Jerry
>
> /*
> * PLcGrid2 is for passing (as arrays of pointers) 2d coordinate
> * transformation arrays.  The grid dimensions are passed for possible
> bounds
> * checking.
> */
>
> typedef struct {
>      PLFLT **xg, **yg, **zg;
>      PLINT nx, ny;
> } PLcGrid2;
>
>
(Replying to self)

Hmmm.... I'll bet zg contains the data to be plotted, xg contains the  
grid points where it "is," and yg contains the grid points where it  
"is supposed to be," and all three are nx by ny in size.

Jerry

[Plplot-general] Question on PlcGrid2

[Jerry <lan...@qw...>]

plot.h contains the following. Is this correct? Is there one extra  
array or one too few dimension sizes or do I need to study the docs  
more?
Jerry

/*
* PLcGrid2 is for passing (as arrays of pointers) 2d coordinate
* transformation arrays.  The grid dimensions are passed for possible  
bounds
* checking.
*/

typedef struct {
     PLFLT **xg, **yg, **zg;
     PLINT nx, ny;
} PLcGrid2;

Re: [Plplot-general] PLplot 5.7.1 is now available

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

On 2006-11-26 22:33-0500 hba...@ma... wrote:

>
> Hello,
>
> PLplot 5.7.1 is now available at Sourceforge. This is the first
> official (development) version of PLplot using the new cmake build
> system. While it can still be built the "old" way (i.e. using
> configure) we recommend getting the latest version of cmake (2.4.4)
> and using that instead. We hope that Windows user in particular will
> have a much easier time installing PLplot. Other changes include (1)
> Modifying PLplot core to buffer plot commands in memory rather than
> via a unix pipe/temporary file for increased speed (2) Numerous
> improvements to the wxWidgets driver and (3) The addition of a new
> driver that creates scalable vector graphics files (SVG).
>

Note, complete instructions for the new CMake Build System (CBS) for PLplot
are given at http://www.miscdebris.net/plplot_wiki/.  This new build system
is much easier for PLplot developers to understand, maintain, and extend.
For both PLplot users and developers our new build system gives more
complete windows builds than our previous (in sys/win32/msdev) home-grown
windows build system, gives excellent results on Linux and Mac OS X, and
gives faster builds than our previous autotools build system (./configure
....).  Since nobody wants to maintain the autotools build system any more
it is now officially deprecated.  We will probably (see Arjen's previous
post about that) deprecate the home-grown windows build system as well.
Therefore, we urge all PLplot users to start using our new CBS to build
PLplot.  Your platform coverage is much more varied than what the developers
have access to so your feedback will be appreciated.

Alan
__________________________
Alan W. Irwin

Astronomical research affiliation with Department of Physics and Astronomy,
University of Victoria (astrowww.phys.uvic.ca).

Programming affiliations with the FreeEOS equation-of-state implementation
for stellar interiors (freeeos.sf.net); PLplot scientific plotting software
package (plplot.org); the Yorick front-end to PLplot (yplot.sf.net); the
Loads of Linux Links project (loll.sf.net); and the Linux Brochure Project
(lbproject.sf.net).
__________________________

Linux-powered Science
__________________________

[Plplot-general] PLplot 5.7.1 is now available

[<hba...@ma...>]

Hello,

PLplot 5.7.1 is now available at Sourceforge. This is the first  
official (development) version of PLplot using the new cmake build  
system. While it can still be built the "old" way (i.e. using  
configure) we recommend getting the latest version of cmake (2.4.4)  
and using that instead. We hope that Windows user in particular will  
have a much easier time installing PLplot. Other changes include (1)  
Modifying PLplot core to buffer plot commands in memory rather than  
via a unix pipe/temporary file for increased speed (2) Numerous  
improvements to the wxWidgets driver and (3) The addition of a new  
driver that creates scalable vector graphics files (SVG).

-Hazen

[Plplot-general] Question about the separate Windows build system

[Arjen M. <arj...@wl...>]

Hello all,

PLplot has several build systems, suited for different platforms:
- The system based on autotools, meant for UNIX, Linux and such platforms
- The system for building the PLplot library on Windows, using the 
native tools
  (compiler, linker, batch files, nmake)

This was necessary, because the autotools do not run on basic Windows
(you need Cygwin or MinGW to get the right environment).

Several months ago, Alan started porting PLplot to the CMake build system
and given the success we have had in adapting it for all platforms we 
actively
support, this "CBS" will be the default build system for PLplot 5.7.0 which
is due in a few days/weeks. The autotools build system will no longer be
supported (or at least not as intensively as it used to).

This raises the question whether the Windows-specific build system should
be deprecated as well. And this is what I wanted to consult you all about.

Let me review the advantages and disadvantages of both systems:

Windows-specific:
- Advantages:
  - Nothing needed except the MSVC 6.0 compiler and the tools that come 
with it
- Disadvantages:
  - Limited support for more modern versions of the MS Visual C/C++ 
compilers
    (you need to build a project for it yourself)
  - Limited support for Fortran (there is a preliminary makefile to 
create the
    Fortran wrapper library), only suitable for Compaq Visual Fortran
 - No support for the scripting (dynamic) languages (Perl, Python and 
Tcl) or
    for Java
 - Currently no support for external libraries, such as GD to make PNG,
   JPEG and GIF files.
 - No support for antialiasing fonts
 - (From a maintainer's point of view) Separate sources that need to be
   adjusted to reflect general development, separate driver (win3) that
   looks a lot like the wingcc driver that is used for Cygwin and MinGW 
under
   Windows.

- Expectations:
  - Further developing this build system will be a time-consuming affair.
    All the nice extra features  (antialiasing fonts, more programming 
languages,
    extra drivers) require considerable updates of the makefiles, as 
there is
    only an ad hoc method to configure them.

CBS:
- Advantages:
  - Generation of makefiles or project files or solutions, depending on your
    preferences and the available tools
  - Support for other C/C++ compilers than MSVC 6.0
  - Support for other builds than the release build (creating a debuggable
    library is easy)
  - Support for Fortran (several compilers) and Tcl.
  - Automatic builds for all sets of examples (C, C++, Fortran, Tcl)
  - (From a maintainer's point of view)
    - The source code is _exactly_ the same as for the UNIX-like 
platforms -
      including the driver (wingcc)
    - It is much easier to expand the supported languages and compilers
      (For Java and Python this has not been done yet. The wrappers are
      created via SWIG and the procedure is well established for 
Linux/UNIX/OSX)

- Disadvantages:
  - You will need to install CMake - that is easy enough, but it is an 
extra step
  - The system is new, it has only been tested by the maintainers and a 
few users
  - There are still a few quirks that need to be solved (this has to do 
with the
    location of the generated DLLs and other output files, but the old 
system
    does not provide much support there either)

- Expectations:
  - With CBS it will be much easier to incorporate developments from the
     other platforms. The main problems have to do with _finding_  the 
various
     external libraries as  under Windows there is no particular 
standard location for
     _libraries_ (as opposed to applications).
  - The deployment component CPack in CMake may be helpful in providing
    binary versions of the libraries. We have no experience with it at 
the moment,
    but it looks like a good solution for people who can not or do not 
want to
    build the library themselves.

So, here are my primary questions:

Is there a need to continue maintaining the old build system, based on 
makefiles
and Windows-specific source code? Would the use of CBS mean that PLplot
becomes inaccessible for you?

I realise that the above description is rather biased towards CBS, but 
it _is_
a much better build system, from a maintainer's point of view at the very
least.

Awaiting your replies,

Arjen

Re: [Plplot-general] Reversing all colors

[Jerry <lan...@qw...>]

On Nov 16, 2006, at 2:23 PM, Alan W. Irwin wrote:

> On 2006-11-16 13:37-0700 Curtis Cooper wrote:
>
>> Hi,
>>
>> For my graphs, it is more convenient to have a white rather than  
>> black
>> background.  However, then my plots don't show up because the axes  
>> boxes,
>> tick marks, etc., are all white by default.  Is it possible to easily
>> reverse all colors in the primary palette?
>
> Both postscript colour devices (psc and psttfc) have monochrome  
> black on
> white counterparts called ps and psttf so you might want to try  
> those device
> drivers.
>
> Also, it is fairly easy to rearrange the discrete colour palette  
> using the
> colour map0 API discussed in
> (http://plplot.sourceforge.net/docbook-manual/plplot-html-5.6.1/ 
> color.html).
>
> In the past, the idea has been discussed on plplot-devel to supply  
> a number
> of convenient pre-defined colour palettes for the user (including the
> default colour palette we have now), but that idea has not been  
> implemented
> yet.
>
> Alan

FWIW, my Ada bindings already have some extra palettes plus the  
ability to switch black and white. I  hope to write Ada versions of  
the PLplot examples (I assume that's a prerequisite) and see what is  
involved with getting this work included with the official release. I  
think there might be a documentation project for the thick binding,  
but the standard PLplot docs should be good for the thin binding.

Jerry

Re: [Plplot-general] Reversing all colors

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

On 2006-11-16 13:37-0700 Curtis Cooper wrote:

> Hi,
>
> For my graphs, it is more convenient to have a white rather than black
> background.  However, then my plots don't show up because the axes boxes,
> tick marks, etc., are all white by default.  Is it possible to easily
> reverse all colors in the primary palette?

Both postscript colour devices (psc and psttfc) have monochrome black on
white counterparts called ps and psttf so you might want to try those device
drivers.

Also, it is fairly easy to rearrange the discrete colour palette using the
colour map0 API discussed in
(http://plplot.sourceforge.net/docbook-manual/plplot-html-5.6.1/color.html).

In the past, the idea has been discussed on plplot-devel to supply a number
of convenient pre-defined colour palettes for the user (including the
default colour palette we have now), but that idea has not been implemented
yet.

Alan
__________________________
Alan W. Irwin

Astronomical research affiliation with Department of Physics and Astronomy,
University of Victoria (astrowww.phys.uvic.ca).

Programming affiliations with the FreeEOS equation-of-state implementation
for stellar interiors (freeeos.sf.net); PLplot scientific plotting software
package (plplot.org); the Yorick front-end to PLplot (yplot.sf.net); the
Loads of Linux Links project (loll.sf.net); and the Linux Brochure Project
(lbproject.sf.net).
__________________________

Linux-powered Science
__________________________

Re: [Plplot-general] Reversing all colors

[Bryan P. <pet...@ma...>]

I have a routine I use all the time in place of plstart which includes two
calls before plstart:

call plscol0(15,0,0,0)
call plscol0(0,255,255,255)

after plstart I also call

call plcol(15)

that swaps the colors.  I assume this still works on the latest version 
(haven't tried it yet).  And I'm not sure that all three calls are needed 
since I have been using this unchanged since late 1997.

Bryan Peterson
bry...@by...

On Thu, 16 Nov 2006, Curtis Cooper wrote:

> Hi,
>
> For my graphs, it is more convenient to have a white rather than black
> background.  However, then my plots don't show up because the axes boxes,
> tick marks, etc., are all white by default.  Is it possible to easily
> reverse all colors in the primary palette?
>
> Thanks,
> Curtis
>
>
> -------------------------------------------------------------------------
> Take Surveys. Earn Cash. Influence the Future of IT
> Join SourceForge.net's Techsay panel and you'll get the chance to share your
> opinions on IT & business topics through brief surveys - and earn cash
> http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV
> _______________________________________________
> Plplot-general mailing list
> Plp...@li...
> https://lists.sourceforge.net/lists/listinfo/plplot-general
>

[Plplot-general] Reversing all colors

[Curtis C. <cu...@lp...>]

Hi,

For my graphs, it is more convenient to have a white rather than black
background.  However, then my plots don't show up because the axes boxes,
tick marks, etc., are all white by default.  Is it possible to easily
reverse all colors in the primary palette?

Thanks,
Curtis

Re: [Plplot-general] cmake installation problem on Win32 platform

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

On 2006-11-03 15:27-0500 Steven Hirshman wrote:

> I download the tar file, expanded it, downloaded the cmake files, installed 
> them.

We haven't (yet) released a tarball with our CMake build system (CBS) work.
So for now you must use the CVS version of PLplot. I have just added that
essential bit of information to http://www.miscdebris.net/plplot_wiki/.
(Sorry we forgot to document this before.) Note, our CVS version of PLplot
works pretty well at the moment with our new CBS, and that work is
accessible to anybody following the directions at our Wiki.  We hope to get
out a development release soon to help publicize our new CBS and also to
provide a fixed version of PLplot for our users to try.

The currently tested platforms are Linux, Mac OS X, Cygwin, MinGW, and bare
windows.  These platforms are essentially all the platforms accessible to
our core developers. More test reports for those platforms and especially
any additional platforms (such as the various traditional Unix variants) are
most welcome.

Alan
__________________________
Alan W. Irwin

Astronomical research affiliation with Department of Physics and Astronomy,
University of Victoria (astrowww.phys.uvic.ca).

Programming affiliations with the FreeEOS equation-of-state implementation
for stellar interiors (freeeos.sf.net); PLplot scientific plotting software
package (plplot.org); the Yorick front-end to PLplot (yplot.sf.net); the
Loads of Linux Links project (loll.sf.net); and the Linux Brochure Project
(lbproject.sf.net).
__________________________

Linux-powered Science
__________________________

[Plplot-general] cmake installation problem on Win32 platform

[Steven H. <hir...@or...>]

I download the tar file, expanded it, downloaded the cmake files, 
installed them.

When I try to run cmake with the incantation:

cmake -G "NMake Makefiles" -DCMAKE_INSTALL_PREFIX=install ..

as told to do on the Wiki site, I receive the error:

"Missing CMakeLists.txt"

Now my understanding is that this is the "Project" file for plplot. 
It was not in the tarred file. Can you tell me where to get it off 
your site, or how to create it so I can generate the plplot library?

I am NOT using cygwin.

Thanks.

Steven P. Hirshman              QPS: Quality, Performance, Science
Fusion Energy Division          hir...@or...
Bldg 5700, Room RF314           Oak Ridge, TN 37830
Tele: 865-574-1289              Fax: 865-576-7770
Yesterday it worked. Today it is not working. Windows is like that.  

Re: [Plplot-general] Possible axis drawing glitch

[Jerry <lan...@qw...>]

On Oct 25, 2006, at 6:51 PM, Alan W. Irwin wrote:

> On 2006-10-25 16:29-0700 Jerry wrote:
>
>> I have encountered a repeatable problem when using my Ada binding  
>> to PLplot,
>> and I'm wondering if it is something that I've done wrong with the  
>> binding.
>
> I suggest the best general approach for debugging a new interface  
> is to
> create or modify a C example to recreate the exact C library  
> conditions that
> are giving you errors for your interface example.  Assuming that the
> C example works and your interface example does not, that very much  
> narrows
> the source of the problem to either the way you wrote your example  
> in the
> interface language or the programming of the interface itself.
>
> In any case, for every new interface we strongly encourage making a  
> complete
> set of standard examples following what is done in examples/c. The
> postscript result from each C example and the corresponding interface
> example should (ideally) be identical (although sometimes there are
> single-digit rounding differences).  The approach of trying to  
> mimic the
> simplest C examples and gradually moving to the more complicated  
> ones often
> helps to isolate interface issues.
>
> Also, you might want to try valgrind.  That is a great tool for  
> figuring out
> memory management issues (if those happen to be the source of some/ 
> all of
> your difficulties).
>
> Alan
> __________________________
> Alan W. Irwin

Problem solved. Operator error. Thanks for your comment.

Jerry

Re: [Plplot-general] Possible axis drawing glitch

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

On 2006-10-25 16:29-0700 Jerry wrote:

> I have encountered a repeatable problem when using my Ada binding to PLplot, 
> and I'm wondering if it is something that I've done wrong with the binding.

I suggest the best general approach for debugging a new interface is to
create or modify a C example to recreate the exact C library conditions that
are giving you errors for your interface example.  Assuming that the
C example works and your interface example does not, that very much narrows
the source of the problem to either the way you wrote your example in the
interface language or the programming of the interface itself.

In any case, for every new interface we strongly encourage making a complete
set of standard examples following what is done in examples/c. The
postscript result from each C example and the corresponding interface
example should (ideally) be identical (although sometimes there are
single-digit rounding differences).  The approach of trying to mimic the
simplest C examples and gradually moving to the more complicated ones often
helps to isolate interface issues.

Also, you might want to try valgrind.  That is a great tool for figuring out
memory management issues (if those happen to be the source of some/all of
your difficulties).

Alan
__________________________
Alan W. Irwin

Astronomical research affiliation with Department of Physics and Astronomy,
University of Victoria (astrowww.phys.uvic.ca).

Programming affiliations with the FreeEOS equation-of-state implementation
for stellar interiors (freeeos.sf.net); PLplot scientific plotting software
package (plplot.org); the Yorick front-end to PLplot (yplot.sf.net); the
Loads of Linux Links project (loll.sf.net); and the Linux Brochure Project
(lbproject.sf.net).
__________________________

Linux-powered Science
__________________________

[Plplot-general] Possible axis drawing glitch

[Jerry <lan...@qw...>]

I have encountered a repeatable problem when using my Ada binding to  
PLplot, and I'm wondering if it is something that I've done wrong  
with the binding. I've narrowed down a specific example, but the  
general problem is (a) the horizontal axis of 2D plots (linear or  
log, at least) is drawn purple, and (b) when there are more than a  
certain number of data points to be drawn, the purple line that  
overlays the horizontal axis swings up to the top of the plot and  
then back down to the axis. (Other glitch-types are possible, but  
this is typical.)

In the specific case, I am drawing a single sine wave. When the  
number of points is 286 or less, there is no glitch. When the number  
of data points is 286 or more, the glitches appear at the right-hand  
side of the plot, with approximately one-half "cycle" of glitches for  
each point over 285. I'm working with OS X and the problem appears  
exactly the same whether using AquaTerm, X11, Postscript color file  
or Postscript monochrome (but obviously without the purple color).

I believe I'm using 5.5.3.

I've attached a PDF which shows the problem when the number of points  
is 297. With thousands of points, the upper part of the plot is  
essentially obscured by purple and some errant lines go downward and  
some even to the edge of the drawing region, beyond the "data part"  
of the plot.

Jerry

Re: [Plplot-general] bug in plplot-config

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

On 2006-10-05 12:07-0700 Joe Chuma wrote:

> Hi,
>
> Installing PLplot-5.6.1 creates the plplot-config and plplot_libtool files
> in ${prefix}/bin. Unfortunately, using `plplot-config --libs` fails with:
> "plplot-config: line 68: plplot_libtool: command not found"
> To fix this, edit plplot-config and change line 68 to:
> `${prefix}/bin/plplot_libtool -n --mode=link ${LINKER} ${plplot_libspec} -o OUTPUT`
> i.e., just add the ${prefix}/bin path to plplot_libtool.
>
> Maybe someone could fix this for future releases?

If your installed bin directory is on your PATH you should not have any
problems since that is where both plplot-config and plplot_libtool are
installed.  So this feature request only helps for the case where you are
invoking plplot-config using the full path, and you have not bothered to put
your installed bin directory on your PATH.

There are some other considerations as well. plplot-config is now deprecated
in favor of our mature pkg-config approach regardless of whether you are
using our autotools or CMake build system.  For example, the only way you
get the plplot-config approach used now when you build our installed
examples is in the unlikely event the user has not installed pkg-config.
Also, plplot-config only works if plplot_libtool is configured and installed
and that will only happen if you use the autotools build system. That build
system will probably be deprecated for our next release in favor of the new
CMake build system.  That new build system is currently only available from
CVS, but it is already looking good enough so I have proposed on
plplot-devel that we make a series of development releases featuring it
leading up to our next stable release where (if all goes well) it will be
the preferred build system.

The proposed fix to plplot-config is simple so we may do it anyway, but I
wanted to take this opportunity to point out how plplot-config and
plplot_libtool are losing importance because of our mature and preferred
pkg-config approach and the good possibility that our new CMake build system
(which necessarily excludes plplot-config and plplot_libtool) will be the
preferred build system for the next release.

Alan
__________________________
Alan W. Irwin

Astronomical research affiliation with Department of Physics and Astronomy,
University of Victoria (astrowww.phys.uvic.ca).

Programming affiliations with the FreeEOS equation-of-state implementation
for stellar interiors (freeeos.sf.net); PLplot scientific plotting software
package (plplot.org); the Yorick front-end to PLplot (yplot.sf.net); the
Loads of Linux Links project (loll.sf.net); and the Linux Brochure Project
(lbproject.sf.net).
__________________________

Linux-powered Science
__________________________

[Plplot-general] bug in plplot-config

[Joe C. <ch...@tr...>]

Hi,

Installing PLplot-5.6.1 creates the plplot-config and plplot_libtool files
in ${prefix}/bin. Unfortunately, using `plplot-config --libs` fails with:
"plplot-config: line 68: plplot_libtool: command not found"
To fix this, edit plplot-config and change line 68 to:
`${prefix}/bin/plplot_libtool -n --mode=link ${LINKER} ${plplot_libspec} -o OUTPUT`
i.e., just add the ${prefix}/bin path to plplot_libtool.

Maybe someone could fix this for future releases?

-- 
Joe Chuma
TRIUMF
Canada's National Laboratory for Particle and Nuclear Physics

Re: [Plplot-general] newbie questions

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

On 2006-09-29 13:02-0400 Steve Friedman wrote:

>
> On Fri, 29 Sep 2006, Alan W. Irwin wrote:

>> You can divide a plot into nxn + 1 rectangular areas (nxn plots +
>> footer) using nxn + 1 calls to plvpor
>
> Is plssub() just a convenience call to plvpor(), or do I need to replicate
> other functionality?

plssub and plvpor are quite different.  plvpor should be all you need to
make nxn +1 rectangular viewports, but you will have to be aware of the
differences between a viewport and a (sub-)page. For a complete discussion
see
http://plplot.sourceforge.net/docbook-manual/plplot-html-5.6.1/viewport_window.html

Alan
__________________________
Alan W. Irwin

Astronomical research affiliation with Department of Physics and Astronomy,
University of Victoria (astrowww.phys.uvic.ca).

Programming affiliations with the FreeEOS equation-of-state implementation
for stellar interiors (freeeos.sf.net); PLplot scientific plotting software
package (plplot.org); the Yorick front-end to PLplot (yplot.sf.net); the
Loads of Linux Links project (loll.sf.net); and the Linux Brochure Project
(lbproject.sf.net).
__________________________

Linux-powered Science
__________________________

Re: [Plplot-general] newbie questions

[Steve F. <st...@ad...>]

On Fri, 29 Sep 2006, Alan W. Irwin wrote:
> On 2006-09-29 09:54-0400 Steve Friedman wrote:
>
>> - I have a set of graphs produced per experimental run that are 4-up pages
>> (e.g., plssub( 2, 2)) and 1-up pages (e.g., plssub(1,1)) and would like to
>> add a common footer to all pages including such things as a description of
>> the experiment and page number.  Neither plmtex() nor plptex() puts text
>> relative to a page, but only relative to the individual viewport.
>> Likewise, on the 4-up pages, I would like to put a header describing the 4
>> graphs.  Ideally, the header/footer should occur prior to plssub() so the
>> plssub() call can take into account the remaining available space.
>
> You can divide a plot into nxn + 1 rectangular areas (nxn plots + 
> footer) using nxn + 1 calls to plvpor

Is plssub() just a convenience call to plvpor(), or do I need to replicate 
other functionality?

> In my opinion we probably don't need a convenience function for the first
> case, but it might be worthwhile to implement trace legends or date (year,
> month, day), and time (hours, minutes, seconds) axes. However, so far we
> have had no C developers volunteer to do such implementations.
>

I know that feeling.

Thanks for the quick reply.

Steve Friedman

Re: [Plplot-general] how to fill a polygon with plplot (pdl interface)

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

> Thanks, Hazen! That's what I was looking for!

> I've found it in "The PLplot Plotting Library / Programmer's
> Reference Manual - by MJ LeBrun & G. Furnish"
> but this manual is C&Fortran oriented and plfill is only descibe with
> the "n" argument which refers to the vertices to be applied in
> the polygon.
>
> Do anyone know a more PDL oriented reference manual? (the description
> of the module in cpan is a bit rough...)

I don't think there is one.  I agree our manual (most up-to-date version is
at http://plplot.sourceforge.net/docbook-manual/plplot-html-5.6.1/) is
currently oriented toward C, fortran 77 (and Tcl/Tk) and needs lots of work
in describing all our other computer language interfaces. Currently our best
documentation of the Fortran 95, Python, Java, and Perl/PDL interfaces to
PLplot is through our set of working examples in the examples sub-directory
of our tarball and also available at
http://plplot.sourceforge.net/examples/index.html.  In general all those
mentioned language interfaces are essentially identical to the
well-documented C interface except that the redundant array dimensions have
been dropped from argument lists (as you can see also from looking at the
examples).

Alan
__________________________
Alan W. Irwin

Astronomical research affiliation with Department of Physics and Astronomy,
University of Victoria (astrowww.phys.uvic.ca).

Programming affiliations with the FreeEOS equation-of-state implementation
for stellar interiors (freeeos.sf.net); PLplot scientific plotting software
package (plplot.org); the Yorick front-end to PLplot (yplot.sf.net); the
Loads of Linux Links project (loll.sf.net); and the Linux Brochure Project
(lbproject.sf.net).
__________________________

Linux-powered Science
__________________________

Re: [Plplot-general] newbie questions

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

On 2006-09-29 09:54-0400 Steve Friedman wrote:

>
> I tried looking through the documentation and examples (BTW:  kudos to
> whoever thought of the examples page), but couldn't locate the commands
> for the following functions.  (For what it is worth, I am running plplot
> 5.6.1 via the python bindings on linux.)
>
> - I have a set of graphs produced per experimental run that are 4-up pages
> (e.g., plssub( 2, 2)) and 1-up pages (e.g., plssub(1,1)) and would like to
> add a common footer to all pages including such things as a description of
> the experiment and page number.  Neither plmtex() nor plptex() puts text
> relative to a page, but only relative to the individual viewport.
> Likewise, on the 4-up pages, I would like to put a header describing the 4
> graphs.  Ideally, the header/footer should occur prior to plssub() so the
> plssub() call can take into account the remaining available space.
>
> - I have a multitrace line plot and would like to add a trace legend.  I
> can't find the convenience function that would draw a line (including the
> current style, color, width, and markers) next to a given label.
>
> - Some of my graphs take time on the x-axis.  However, it seems that an
> array of strings is not accepted for the labels.  It seems that my
> alternatives are either to use fractions of hours or call plmtex() myself.
> Am I missing anything?

Sorry, but there are no convenience functions for any of these cases.
However, there are alternatives. You can divide a plot into nxn + 1
rectangular areas (nxn plots + footer) using nxn + 1 calls to plvpor. It
should be straightforward to combine low-level function calls to implement
the trace legend that you have described. Finally, to deal with date or time
plots, a reasonable fallback is to use fractions of a year, day, or hour or
else use the old astronomical research trick of plotting time coordinates in
Julian days minus some convenient offset.

In my opinion we probably don't need a convenience function for the first
case, but it might be worthwhile to implement trace legends or date (year,
month, day), and time (hours, minutes, seconds) axes. However, so far we
have had no C developers volunteer to do such implementations.

Alan
__________________________
Alan W. Irwin

Astronomical research affiliation with Department of Physics and Astronomy,
University of Victoria (astrowww.phys.uvic.ca).

Programming affiliations with the FreeEOS equation-of-state implementation
for stellar interiors (freeeos.sf.net); PLplot scientific plotting software
package (plplot.org); the Yorick front-end to PLplot (yplot.sf.net); the
Loads of Linux Links project (loll.sf.net); and the Linux Brochure Project
(lbproject.sf.net).
__________________________

Linux-powered Science
__________________________

[Plplot-general] newbie questions

[Steve F. <st...@ad...>]

I tried looking through the documentation and examples (BTW:  kudos to 
whoever thought of the examples page), but couldn't locate the commands 
for the following functions.  (For what it is worth, I am running plplot 
5.6.1 via the python bindings on linux.)

- I have a set of graphs produced per experimental run that are 4-up pages 
(e.g., plssub( 2, 2)) and 1-up pages (e.g., plssub(1,1)) and would like to 
add a common footer to all pages including such things as a description of 
the experiment and page number.  Neither plmtex() nor plptex() puts text 
relative to a page, but only relative to the individual viewport. 
Likewise, on the 4-up pages, I would like to put a header describing the 4 
graphs.  Ideally, the header/footer should occur prior to plssub() so the 
plssub() call can take into account the remaining available space.

- I have a multitrace line plot and would like to add a trace legend.  I 
can't find the convenience function that would draw a line (including the 
current style, color, width, and markers) next to a given label.

- Some of my graphs take time on the x-axis.  However, it seems that an 
array of strings is not accepted for the labels.  It seems that my 
alternatives are either to use fractions of hours or call plmtex() myself. 
Am I missing anything?

Steve Friedman

Re: [Plplot-general] how to fill a polygon with plplot (pdl interface)

[Bruno P. <bru...@cl...>]

<!DOCTYPE html PUBLIC "-//W3C//DTD HTML 4.01 Transitional//EN">
<html>
<head>
  <meta content="text/html;charset=ISO-8859-1" http-equiv="Content-Type">
  <title></title>
</head>
<body bgcolor="#ffffff" text="#000000">
Thanks, Hazen! That's what I was looking for!<br>
<br>
I've found it in "The PLplot Plotting Library / Programmer's Reference
Manual - by MJ LeBrun &amp; G. Furnish"<br>
but this manual is C&amp;Fortran oriented and plfill is only descibe
with the "n" argument which refers to the vertices to be applied in<br>
the polygon.<br>
<br>
Do anyone know a more PDL oriented reference manual? (the description
of the module in cpan is a bit rough...)<br>
<br>
best,<br>
<br>
Bruno<br>
<br>
<a class="moz-txt-link-abbreviated" href="mailto:hba...@ma...">hba...@ma...</a> a &eacute;crit&nbsp;:
<blockquote cite="mid...@ma..."
 type="cite"><br>
On Sep 28, 2006, at 5:47 AM, Bruno Picard wrote:
  <br>
  <br>
  <blockquote type="cite">&nbsp;Fast:
    <br>
Is there a plplot equivalent to the matlab function FILL?:
    <br>
FILL Filled 2-D polygons.
    <br>
&nbsp;&nbsp;&nbsp; FILL(X,Y,C) fills the 2-D polygon defined by vectors X and Y
    <br>
&nbsp;&nbsp;&nbsp; with the color specified by C.&nbsp; The vertices of the polygon
    <br>
&nbsp;&nbsp;&nbsp; are specified by pairs of components of X and Y.&nbsp; If necessary,
    <br>
&nbsp;&nbsp;&nbsp; the polygon is closed by connecting the last vertex to the first.
    <br>
  </blockquote>
  <br>
I believe that the function you seek is called plfill.
  <br>
  <br>
An example using the PDL interface:
  <br>
  <br>
#!/usr/bin/perl -w
  <br>
  <br>
use strict;
  <br>
use PDL;
  <br>
use PDL::Graphics::PLplot;
  <br>
  <br>
plspage(0, 0, 300, 300, 0, 0);
  <br>
plinit();
  <br>
pladv(0);
  <br>
plvsta();
  <br>
plwind(0.0, 4.0, 0.0, 4.0);
  <br>
plbox(0.0, 0, 0.0, 0, "bcnst", "bcnstv");
  <br>
  <br>
plcol0(2);
  <br>
my $x = pdl(1.0, 1.0, 2.0, 2.5);
  <br>
my $y = pdl(1.0, 2.2, 2.1, 1.0);
  <br>
plfill($x, $y);
  <br>
  <br>
plend();
  <br>
  <br>
  <br>
The polygon is automatically closed and is drawn with the current
color. Also, note that the PDL interface automatically fills in the "n"
in plfill(n,x,y).
  <br>
  <br>
best,
  <br>
-Hazen
  <br>
  <br>
  <br>
  <br>
  <br>
  <br>
</blockquote>
<br>
<div class="moz-signature">-- <br>
<div align="center"><u>____________________________________________________</u>
<br>
<font face="Verdana" size="2">Dr. Bruno Picard<br>
D&eacute;p. Traitement de la Mesure et Segment Sol <br>
(Data Analysis and Ground Processing Unit)
&nbsp;<br>
CLS<br>
Dir.&nbsp; Oc&eacute;anographie Spatiale<br>
8-10 rue Herm&egrave;s, 31520 Ramonville Saint Agne, France<br>
<br>
T&eacute;l: (+33)5.61.39.37.37, Fax: (+33)5.61.39.37.82<br>
M&eacute;l: <a href="mailto:bru...@cl...">bru...@cl...</a><br>
Toile : <a href="http://julie-et-bruno.9online.fr/nono/cadrenono.html"
 eudora="autourl">http://julie-et-bruno.9online.fr/nono/cadrenono.html</a><br>
</font><u>____________________________________________________</u><br>
<img src="cid:par...@cl..."><br>
</div>
</div>
</body>
</html>

Re: [Plplot-general] how to fill a polygon with plplot (pdl interface)

[<hba...@ma...>]

On Sep 28, 2006, at 5:47 AM, Bruno Picard wrote:

>  Fast:
> Is there a plplot equivalent to the matlab function FILL?:
> FILL Filled 2-D polygons.
>     FILL(X,Y,C) fills the 2-D polygon defined by vectors X and Y
>     with the color specified by C.  The vertices of the polygon
>     are specified by pairs of components of X and Y.  If necessary,
>     the polygon is closed by connecting the last vertex to the first.

I believe that the function you seek is called plfill.

An example using the PDL interface:

#!/usr/bin/perl -w

use strict;
use PDL;
use PDL::Graphics::PLplot;

plspage(0, 0, 300, 300, 0, 0);
plinit();
pladv(0);
plvsta();
plwind(0.0, 4.0, 0.0, 4.0);
plbox(0.0, 0, 0.0, 0, "bcnst", "bcnstv");

plcol0(2);
my $x = pdl(1.0, 1.0, 2.0, 2.5);
my $y = pdl(1.0, 2.2, 2.1, 1.0);
plfill($x, $y);

plend();


The polygon is automatically closed and is drawn with the current  
color. Also, note that the PDL interface automatically fills in the  
"n" in plfill(n,x,y).

best,
-Hazen

[Plplot-general] how to fill a polygon with plplot (pdl interface)

[Bruno P. <bru...@cl...>]

<!DOCTYPE html PUBLIC "-//W3C//DTD HTML 4.01 Transitional//EN">
<html>
<head>
  <meta content="text/html;charset=ISO-8859-1" http-equiv="Content-Type">
</head>
<body bgcolor="#ffffff" text="#000000">
Hi all,<br>
<br>
Is plplot a real equivalent to matlab? of course not, it is better!<br>
But, well, am I in front of an issue that may change my mind?<br>
NO, it's impossible!! so please, help me!<br>
<br>
Fast:<br>
Is there a plplot equivalent to the matlab function FILL?:<br>
FILL Filled 2-D polygons.<br>
&nbsp;&nbsp;&nbsp; FILL(X,Y,C) fills the 2-D polygon defined by vectors X and Y<br>
&nbsp;&nbsp;&nbsp; with the color specified by C.&nbsp; The vertices of the polygon<br>
&nbsp;&nbsp;&nbsp; are specified by pairs of components of X and Y.&nbsp; If necessary,<br>
&nbsp;&nbsp;&nbsp; the polygon is closed by connecting the last vertex to the first.<br>
<br>
Detailled:<br>
I'm using PDL plplot interface and i would like to plot an image pixel
by pixel<br>
(the pixels of this image are rectangular and have different sizes).<br>
I've read plplot doc, have a look to x15.pl and x16.pl examples but,
well, :-(<br>
I have not understood anaything.<br>
Anyway, I've found a VERY INEFFICIENT way to do it:<br>
$z is my image, (Xpos* and Ypos*) are the limits of each pixels<br>
<br>
<tt>for ($k=0;$k&lt;$Nbpix;$k++)<br>
{<br>
&nbsp;&nbsp;&nbsp; $w = ones(2,2)*$z($k);<br>
&nbsp;&nbsp;&nbsp; plshades ($w, <br>
&nbsp;&nbsp;&nbsp; &nbsp;&nbsp;&nbsp; &nbsp; $XposLeft($k),<br>
&nbsp;&nbsp;&nbsp; &nbsp;&nbsp;&nbsp; &nbsp; $XposRight($k),<br>
&nbsp;&nbsp;&nbsp; &nbsp;&nbsp;&nbsp; &nbsp; $YposBot($k),&nbsp;&nbsp;&nbsp; &nbsp;&nbsp;&nbsp;&nbsp;&nbsp; <br>
&nbsp;&nbsp;&nbsp; &nbsp;&nbsp;&nbsp; &nbsp; $YposUp($k),&nbsp;&nbsp;&nbsp; &nbsp;&nbsp;&nbsp;&nbsp;&nbsp; &nbsp;&nbsp;&nbsp; &nbsp;&nbsp;&nbsp; &nbsp;&nbsp;&nbsp;&nbsp;&nbsp; <br>
&nbsp;&nbsp;&nbsp; &nbsp;&nbsp;&nbsp; &nbsp; $shedge, $fill_width,<br>
&nbsp;&nbsp;&nbsp; &nbsp;&nbsp;&nbsp; &nbsp; $cont_color, $cont_width, 1, 0, 0, 0);<br>
}<br>
</tt>Anyway, since $Nbpix is 512*100, the display takes a long time....<br>
Perhaps there is a way with \&amp;pltr(1or2) and $cgrid(1or2) as in the
examples<br>
but i did not understand the impact on the surface....<br>
<br>
thanks all,<br>
<br>
best regards,<br>
<br>
bruno<br>
<br>
<br>
<div class="moz-signature">-- <br>
<div align="center"><u>____________________________________________________</u>
<br>
<font face="Verdana" size="2">Dr. Bruno Picard<br>
D&eacute;p. Traitement de la Mesure et Segment Sol <br>
(Data Analysis and Ground Processing Unit)
&nbsp;<br>
CLS<br>
Dir.&nbsp; Oc&eacute;anographie Spatiale<br>
8-10 rue Herm&egrave;s, 31520 Ramonville Saint Agne, France<br>
<br>
T&eacute;l: (+33)5.61.39.37.37, Fax: (+33)5.61.39.37.82<br>
M&eacute;l: <a href="mailto:bru...@cl...">bru...@cl...</a><br>
Toile : <a href="http://julie-et-bruno.9online.fr/nono/cadrenono.html"
 eudora="autourl">http://julie-et-bruno.9online.fr/nono/cadrenono.html</a><br>
</font><u>____________________________________________________</u><br>
<img src="cid:par...@cl..."><br>
</div>
</div>
</body>
</html>

[Plplot-general] A completely static build of a PLplot example

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

On 2006-09-21 18:44-0700 Alan W. Irwin wrote:

> The current status is the CVS version of PLplot should give a static build
> of PLplot libraries (including the C++ wxwidgets and psttf devices) that
> passes all build-tree tests, but there is still a minor problem (library
> ordering issue) for the install tree tests which I will deal with tomorrow.

That problem is now sorted out for our new CBS.

>
> Note, this PLplot "static build" is still not what Joe wants since even
> though the PLplot libraries are all statically linked together for this
> case, the external linking is still shared.  I think the solution to making
> even the external linking shared is simply to impose the -Bstatic link
> option on top of the static PLplot build (once that completely works), but I
> haven't tried that yet, and I am sure there will be minor but time-consuming
> issues to sort out.

It turned out there were a lot of those, but I finally got a completely
static build (with our new CBS) of one of the installed examples to work
with the wxwidgets device driver.

So Joe, here is the cookbook you should follow if you want a completely
static build (no external library dependencies at all) of an executable that
uses our plplot library (with embedded wxwidgets device and static linking
to all the wxwidgets related libraries). If there is any part of the
cookbook you do not understand, contact me off list.

(1) Build a static version of wxwidgets.  I don't think any distribution
does that because there is a choice of static or shared, but apparently not
both. Certainly, Debian and derivative like Ubuntu do not provide a static
version of the wxwidgets libraries.

I used wxGTK-2.6.3.tar.gz from SourceForge, and did their recommended
out-of-source build using --disable-shared (which is their option for a
static build).  I also did a shared build as well because that is what I
normally use, but I used a unique build tree and installation prefix for
that so the shared and static results could not clash with each other.

(2) Build PLplot with our new CBS.  On your Linux build system follow the
general Unix directions on our wiki (http://www.miscdebris.net/plplot_wiki).
(Those instructions include downloading and building the latest version of
cmake. That is currently 2.4.3.  Don't use a lower version than that.)

(2a) Preparation for the build using CMake.

The wiki discusses helping cmake to find system resources.  In my case I
installed the above static wxwidgets package with a unique prefix of
/usr/local/wxwidgets.  Thus, I had to set

PATH="/usr/local/wxwidgets/bin:"$PATH

That allowed cmake to find the correct (static) version of wx-config and
bootstrap all wxwidgets information from there.

Here is the cmake command-line I used to pare everything down to a minimal
PLplot system with one (wxwidgets) device:

cmake -DCMAKE_INSTALL_PREFIX=/home/software/plplot_cvs/installcmake \
-DCMAKE_VERBOSE_MAKEFILE=ON -DENABLE_cxx=OFF -DENABLE_f77=OFF \
-DENABLEf95=OFF -DENABLE_python=OFF -DENABLE_java=OFF -DENABLE_tcl=OFF \
-DDEFAULT_NO_DEVICES=ON -DPLD_wxwidgets=ON -DBUILD_SHARED_LIBS=OFF \
../plplot_cmake >& cmake.out

You will undoubtedly want to use your own name for the installation prefix
directory location and for plplot_cmake (a freshly checked out PLplot source
tree from SourceForge). The rest of your cmake options should be identical
with those above. Those options assure verbose make output (including all
the compiler commands), turn off all language interfaces to our core C
library, turn off all devices other than wxwidgets, and build plplot
internally with static libraries.

(2b) The build.  Stay in the separate build tree you prepared with the
above cmake command and invoke

make >& make.out

You will find it doesn't work because you are mixing C and C++ code
(including the static wxwidgets library code as well as the wxwidgets PLplot
device code) into all executables you build that refer to the C++ wxwidgets
device.  We only have one such executable (an executable called plrender).
One method of beating the linking issue is to find the exact gcc command
that did not work from the above make.out file, and replace gcc with g++ to
get plrender linked properly for this special case of a static wxwidgets
library. However, for your minimal needs you don't really need plrender so
just touch the appropriate plrender file and proceed with make again to
finish the build.

(2c) The install.  Stay in the same directory and invoke

make install >& make_install.out

Again, there is trouble with plrender, but you solve it the same way. 
"touch" is easier than invoking an exact g++ command.  However, be sure to
touch plrender in the correct directory given by make_install.out rather
than the different directory where you touched plrender in (2b) above.
Do "make install" after the touch to finish the install.

Here are the essential installation results you should get from the above
steps.

cd /home/software/plplot_cvs/installcmake
find -type f
....
./share/plplot5.6.1/plstnd5.fnt
./share/plplot5.6.1/plxtnd5.fnt
....
./share/plplot5.6.1/examples/Makefile
./share/plplot5.6.1/examples/c/plcdemos.h
./share/plplot5.6.1/examples/c/x01c.c
./share/plplot5.6.1/examples/c/Makefile
....
./lib/libcsirocsa.a
./lib/libcsironn.a
./lib/libplplotd.a
./lib/pkgconfig/plplotd.pc
....
./include/plplot/plplot.h
+ many more essential PLplot headers.

(2d) Build the x01c example for static PLplot and wxwidgets libraries, but
with everything else shared.

cd /tmp
cp -a /home/software/plplot_cvs/installcmake/share/plplot5.6.1/examples .
cd examples/c
make x01c >& make_x01c.out

Again it won't work because of C/C++ linking issues and this time really
do replace gcc with g++ to fix the problem.  Note, this result does
use the static PLPlot libraries (and you should check that ./x01c -dev
wxwidgets gives good results) but we need one more step to make
that x01c executable completely static.

(2e) Build a completely static x01c example.

The above command for (2d) is transformed to a completely static build
with the -static gcc option.  (Also, you don't need rpath
since that is not relevant for a completly static build).  Also, you have
to add some libraries. Static linking must mention all the non-system
libraries in the correct order and in general you cannot be as sloppy about
libraries as in the shared case.  The result is as follows:

software@chickadee> g++ x01c.c -o x01c_static -static \
`PKG_CONFIG_PATH=/home/software/plplot_cvs/installcmake/lib/pkgconfig pkg-config --cflags --libs plplotd` \
-lX11 -lfontconfig -lXcursor -lXrender \
-lXext -lXft -lXrandr -lfreetype -lfontconfig -lX11 -lXi -lexpat -lXrender \
-lpangoft2-1.0 2>&1 |grep undefined |sort |uniq|wc -l
0

Note, all the required libraries that are listed out with the pkg-config
command supplemented by the trailing -l options must have the *.a static
form of library installed.  It was at this stage I discovered I would have
to build my own static version of the wxwidgets library since Ubuntu (and
Debian) do no supply the static version.  It's possible you will have
similar trouble with the other required libraries as well.  The solution is
usually to install -devel versions of library packages.

I used the grep undefined trick to keep track of all the undefined symbols
you get with static linking if you don't take care to specify all libraries
The wx-config result for wxwidgets (which is what our CBS uses to configure
the above pkg-config command) is good enough for shared linking, but doesn't
cut the mustard for static linking which is why I had to specify all those
additional libraries with the above -l options.  You also have to be careful
of static library order which is why some of those -l options are repeated.
The 0 at the end showed there were (finally after several hours effort to
correlate every undefined symbol with a library using google searches and
sometimes guesses) no undefined symbols in the result.

(3) Results:

Here is the size of the result:

software@chickadee> ls -lh x01c_static
-rwxr-xr-x  1 software software 12M Sep 22 22:38 x01c_static*

(I was actually expecting bigger so I am pleasantly surprised by that.)

ldd confirms the resulting x01c_static is completely static with no
dependencies on libraries.

./x01c_static gives the same results as the shared version (x01c built in step
(2d)

You must distribute the Hershey fonts with your application (plstnd5.fnt and
plxtnd5.fnt from above).  The location of those fonts can be specified with
the PLPLOT_LIB environment variable and if that is not specified or that
location does not contain the Hershey fonts, then it falls back to the
hard-coded install location, $prefix/share/plplot5.6.1 on the build box, see
above. Thus, your users will probably have to set PLPLOT_LIB since the
hard-coded location is unlikely to be relevant on their machine.

The above fonts are only 65K so it is not going to be onerous for you to
distribute them in addition to the large executable that is obtained from a
static build.

Note, one downside of the Hershey fonts are the resulting characters are
crude (completely unhinted) for screen display.  Fortunatley, the wxwidgets
device can also use a high-quality font alternative.  I have confirmed that
alternative font system works well with ./x01c_static, but you should cross
that bridge when you come to it.  For now, please concentrate on getting the
static x01c example built and working with the default Hershey fonts
following the instructions above.

This has been an interesting project, but I am also glad to be done with it.
For a while there I thought I was in a time warp to the early 80's when
every developer had to contend with static linking library order problems
and trying to figure out which undefined symbol came from what library.  In
those days you looked everything up in a large room completely filled with
computer manuals.  Now it is significantly easier with google, but still
difficult.

Good luck in mimicking my completely static build result.

Alan
__________________________
Alan W. Irwin

Astronomical research affiliation with Department of Physics and Astronomy,
University of Victoria (astrowww.phys.uvic.ca).

Programming affiliations with the FreeEOS equation-of-state implementation
for stellar interiors (freeeos.sf.net); PLplot scientific plotting software
package (plplot.org); the Yorick front-end to PLplot (yplot.sf.net); the
Loads of Linux Links project (loll.sf.net); and the Linux Brochure Project
(lbproject.sf.net).
__________________________

Linux-powered Science
__________________________