gnuplot-beta — Developers and beta testers of new gnuplot code

Put ps_fontfile_doc.pdf on gnuplot web page

[Harald H. <har...@gm...>]

Dear all,

maybe, this mailing list is the wrong address for this question, but I 
don't know better, at the moment.

Today, somebody requested a link to the file ps_fontfile_doc.pdf (Using 
TeX Fonts in the Gnuplot Postscript Terminal) in the WWW which is part of 
the gnuplot distribution. Unfortunately, it is not accessable via the WWW 
pages below http://gnuplot.sourceforge.net/. I would appreciate if the 
document could be made downloadable either on the "Documentation" or the 
"Tutorials, learning, help" page.

Could you please let me know what you think about that?

Best regards
Harald

-- 
Harald Harders
h.h...@tu...

Re: a minor update to makefile.mgw: Help Workshop

[Mojca M. <moj...@gm...>]

On 2/24/07, Hans-Bernhard Br=F6ker wrote:
> Mojca Miklavec wrote:
> > windows, I found out that the web reference to Help Workshop mentioned
> > in the makefile isn't working any more (I guess that the domain has
> > changed).
>
> I'm afraid the same would apply to any change we make to this
> information, rather sooner than later.  Microsoft has made a sport out
> of completely rearranging their websites on a regular basis, meaning no
> deep link into it is likely to remain available for any useful amount of
> time.  The only stable location specifier would be "somewhere on the
> net, probably at Microsoft".

I don't visit Microsoft pages very often, but hardcoding the link
indeed doesn't seem too reliable.

But it makes no sense to keep broken/non-existing links in comments
either. I would mention that one should probably look for
"hcwsetup.exe" on Microsoft pages (in addition to what is already
written). It's not evident from makefile.mgw whether that is a
Microsoft product or some third-party tool. I wouldn't be surprised if
some tool on some X page would promote that name, but contain a virus
instead.

But it's only a minor (unimportant) change after all, no need to loose
time on it.

Mojca

Re: Let's get 4.2 out as it is. Re: [ gnuplot-Bugs-1655001 ] set grid front overwrites zeroaxes

[Daniel J S. <dan...@ie...>]

Ethan A Merritt wrote:
> On Saturday 24 February 2007 02:04, Petr Mikulik wrote:
> 
>>Bugs item #1655001, was opened at 2007-02-08 10:18
>>You can respond by visiting: 
>>https://sourceforge.net/tracker/?func=detail&atid=102055&aid=1655001&group_id=2055
>>
>>Ethan, now this works differently in gnuplot 4.2rc4 and 4.3cvs:
>>
>>Is it intended? Could you please apply your patch for 4.3 also for 4.2?
> 
> 
> I do not want to make any more code changes for 4.2.
> Last minute configuration fixes or documentation tweaks only.
> Let's get it out the door!

Agreed.

> This patch ("splot_map_border") is one of a dozen or so that I
> have been collecting against 4.2.
> Each is self-contained and already in cvs, but was developed
> after the 4.2 code freeze. I give a complete list below.
> 
> My thought was to bundle these with fixes for any serious bugs
> reported for the distributed 4.2 code, and make them available
> several months after 4.2 goes out as "patch level 1".

If you want, but will patch levels get the consideration that a new release does to justify the effort?  (I'm not sure how software bundles are maintained.)  Otherwise, use your judgement of what to move into the release.  If you want help verifying behavior of any of these, just make a  short comment of what command or demo to look at for each:

>  1) rgbvar2D              Allow 3rd col of color info in 2D plots
>  2) image_clip            Implements image clipping in post pdf wxt gd
>  3) pdf_fonts             error trap and recovery for missing font info
>  4) offset_ambiguity      reduce the problem of trailing comma after 2D offset
>  5) 3dimpulses            splot with impulses lt rgb variable
>  6) zero_width_sample     if sample box is 0 width, don't draw boundary either
>  7) svg_pathlength        avoid ridiculously long lines in svg output
>  8) clipping_failure      fixes bug #1649887 that triggered a segfault when
>                           line segment endpoint went negative in bitmap.c
>  9) splot_map_border      Force border in 'set view map' to be either front or back
> 10) gd_grid               Draw dotted grid as alternating current_color/transparent
> 11) label_def_pt          Provide a default point type for labels
> 12) int_error_reset       Call back into datafile.c from int_error() to reset context
> 13) strcol                Allow strcol() as shorthand for stringcolumn()

Dan

Let's get 4.2 out as it is. Re: [ gnuplot-Bugs-1655001 ] set grid front overwrites zeroaxes

[Ethan A M. <merritt@u.washington.edu>]

On Saturday 24 February 2007 02:04, Petr Mikulik wrote:
> Bugs item #1655001, was opened at 2007-02-08 10:18
> You can respond by visiting: 
> https://sourceforge.net/tracker/?func=detail&atid=102055&aid=1655001&group_id=2055
> 
> Ethan, now this works differently in gnuplot 4.2rc4 and 4.3cvs:
> 
> Is it intended? Could you please apply your patch for 4.3 also for 4.2?

I do not want to make any more code changes for 4.2.
Last minute configuration fixes or documentation tweaks only.
Let's get it out the door!

This patch ("splot_map_border") is one of a dozen or so that I
have been collecting against 4.2.
Each is self-contained and already in cvs, but was developed
after the 4.2 code freeze. I give a complete list below.

My thought was to bundle these with fixes for any serious bugs
reported for the distributed 4.2 code, and make them available
several months after 4.2 goes out as "patch level 1".

 1) rgbvar2D              Allow 3rd col of color info in 2D plots
 2) image_clip            Implements image clipping in post pdf wxt gd
 3) pdf_fonts             error trap and recovery for missing font info
 4) offset_ambiguity      reduce the problem of trailing comma after 2D offset
 5) 3dimpulses            splot with impulses lt rgb variable
 6) zero_width_sample     if sample box is 0 width, don't draw boundary either
 7) svg_pathlength        avoid ridiculously long lines in svg output
 8) clipping_failure      fixes bug #1649887 that triggered a segfault when
                          line segment endpoint went negative in bitmap.c
 9) splot_map_border      Force border in 'set view map' to be either front or back
10) gd_grid               Draw dotted grid as alternating current_color/transparent
11) label_def_pt          Provide a default point type for labels
12) int_error_reset       Call back into datafile.c from int_error() to reset context
13) strcol                Allow strcol() as shorthand for stringcolumn()


-- 
Ethan A Merritt
Biomolecular Structure Center
University of Washington, Seattle 98195-7742

Re: a minor update to makefile.mgw: Help Workshop

[<HBB...@t-...>]

Mojca Miklavec wrote:
> windows, I found out that the web reference to Help Workshop mentioned
> in the makefile isn't working any more (I guess that the domain has
> changed). 

I'm afraid the same would apply to any change we make to this 
information, rather sooner than later.  Microsoft has made a sport out 
of completely rearranging their websites on a regular basis, meaning no 
deep link into it is likely to remain available for any useful amount of 
time.  The only stable location specifier would be "somewhere on the 
net, probably at Microsoft".

a minor update to makefile.mgw: Help Workshop

[Mojca M. <moj...@gm...>]

Hello,

during my failures to try to compile the gnuplot binary properly on
windows, I found out that the web reference to Help Workshop mentioned
in the makefile isn't working any more (I guess that the domain has
changed). It would be probably more helpful to mention that one should
try to download hcwsetup.exe from Microsoft web pages. (google for
"hcwsetup.exe site:microsoft.com")

Here's one page from where it can be downloaded, although I'm not 100%
sure that the link to it is stable:
http://download.microsoft.com/download/win95/MSPRESS/1.0/W9X/EN-US/hcwsetup.EXE

And some useful description:
http://www.microsoft.com/technet/prodtechnol/office/office2000/proddocs/opg/part2/ch13.mspx

Mojca

Re: www.gnuplot.info

[Petr M. <mi...@ph...>]

> 
> > Into ftp.gnuplot.info: there is a faq/ directory, I propose to remove
> >    it, or write "go to web"
> 
> This raises the question: do we really still need ftp access? I like the
> legacy files (notably the gnuplot-historical), but we could also move
> those to the web repository. I haven't checked about disk space, so
> maybe we shouldn't do that, but I wonder whether we really still need
> ftp access as a protocol.

I've copied the contents of ftp.gnuplot.info into
	http://gnuplot.sourceforge.net/ftp_testing/ftp_gnuplot_info/

and made some structural changes. Please continue.

There are also other dirs on ftp.gnuplot.info:
- gnuplot-historical: could be moved inside the above
- gnuplot-misc could be removed
- gnuplot-recent ???
- winsock can be removed

I think ftp.gnuplot.info should be kept, as it is mirrored by CTAN and maybe 
some other repositories.

---
PM

Thank You

[BBands <bb...@ya...>]

This is a spontaneous appreciation after an afternoon spent working with
gnuplot and gnuplot.py.

Simply a superb effort!

Thank you,

     jab


John Bollinger, CFA, CMT
www.BollingerBands.com

If you advance far enough, you arrive at the beginning.


 
____________________________________________________________________________________
Any questions? Get answers on any topic at www.Answers.yahoo.com.  Try it now.

Re: [Gnuplot-bugs] changing color scale with pm3d bug report

[<HBB...@t-...>]

David Koes wrote:
> If I do the following:
> set pm3d
> splot "foo" matrix
> 
> where foo contains:
> 0.492346689417028	0.492346705143181	0.488775972023425	 
> 0.488776412355709	0.488414794709724	0.488413961223616	0.488416521965528	

... then you're putting a very strange task to gnuplot.  You claim you 
have a matrix-style datafile, but actually you don't.

> the plot displays fine, 

Then what you showed can't be what you actually tested, I think.  The 
commands and data you sent consistently produce nothing but an error 
message: "Matrix does not represent a grid" here, both for version 4.0 
and the 4.2 branch head.

changing color scale with pm3d bug report

[David K. <ko...@cm...>]

Hi,

If I do the following:
set pm3d
splot "foo" matrix

where foo contains:
0.492346689417028	0.492346705143181	0.488775972023425	 
0.488776412355709	0.488414794709724	0.488413961223616	0.488416521965528	
0.492406275810712	0.492346762805742	0.488775240757311	 
0.488775293177821	0.489158712513903	0.488414372724619	0.488415688479419	
0.49234515087506	0.492345074865321	0.4887754451973	0.488775217168082	 
0.488775185715776	0.48915856311545	0.488414102758992	
0.492347027529317	0.492345274063259	0.488773589511247	 
0.488775253862439	0.488775243378337	0.488775185715776	0.489156864690927	
0.492346791637023	0.492345468019146	0.488774323398387	 
0.488775177852699	0.488775243378337	0.488775298419872	0.488775322009102	
0.492346720869334	0.49234677591087	0.488774360092744	 
0.488773468944074	0.488775243378337	0.488775175231674	0.488775196199878	
0.49234682833138	0.492346715627283	0.488775293177821	 
0.488773791330211	0.488775180473725	0.488775128053215	0.488775272209617	


the plot displays fine, but if I change the view interactively, the  
color scale changes (to be less useful).  The plot is still accurate,  
but the color scale changes so that
the whole plot is basically the same color.

I've confirmed this problem on Redhat 9 with gnuplot 4.2rc4 and  
Fedora Core 6 with 4.2rc4 and 4.0.  Interestingly, it isn't a problem  
on OS X using the X11 terminal and 4.2rc4.

Thanks,
-Dave

Re: Interesting problem - date arithmetics, compariosn

[Lars H. <lhe...@us...>]

Ethan A Merritt writes:
> On Friday 23 February 2007 07:41, Lars Hecking wrote:
> > 
> >  Similarily, I'd like to define a function based on time/date values. E.g.
> > 
> >  f(x) = ( x < Jan 01 2007 13:00  ? 30 : 40 )
> > 
> >  Any ideas how to achieve this? Unless I've missed some documentation, this
> >  is probably a nice little project for the post-4.2 era  :)
> 
> There has been a patchset from Juergen Wieferink on SourceForge for
> more than a year that extends the options available for handling dates.
> [ 1199186 ] strftime and strptime
 
 I'll check it out, thanks!

> I have to admit that I never looked at it.  Date/time handling is just not
> a part of gnuplot that I have ever needed to use.
 
 You're not a sysadmin then ;-)

Re: Interesting problem - date arithmetics, compariosn

[Ethan A M. <merritt@u.washington.edu>]

On Friday 23 February 2007 07:41, Lars Hecking wrote:
> 
>  Similarily, I'd like to define a function based on time/date values. E.g.
> 
>  f(x) = ( x < Jan 01 2007 13:00  ? 30 : 40 )
> 
>  Any ideas how to achieve this? Unless I've missed some documentation, this
>  is probably a nice little project for the post-4.2 era  :)

There has been a patchset from Juergen Wieferink on SourceForge for
more than a year that extends the options available for handling dates.
[ 1199186 ] strftime and strptime

I have to admit that I never looked at it.  Date/time handling is just not
a part of gnuplot that I have ever needed to use.


-- 
Ethan A Merritt
Biomolecular Structure Center
University of Washington, Seattle 98195-7742

Interesting problem - date arithmetics, compariosn

[Lars H. <lhe...@us...>]

 I came across an interesting problem, and have no (easy) solution for it:

 I would like to plot a datafile with time/date information

date val1 val2 ...
date val1 val2 ...
date val1 val2 ...

 but iwant to plot only certain dates. Via either "using" or the ternary
 operator, I need to compare a date against a given date and take action
 depending on the result. E.g. if the date from the data file is before
 Jan 01 2007, don't plot. In simple cases, and if the input can be
 controlled accordingly, this could be done via data index, but not
 necessarily in more complex cases.

 Similarily, I'd like to define a function based on time/date values. E.g.

 f(x) = ( x < Jan 01 2007 13:00  ? 30 : 40 )

 Any ideas how to achieve this? Unless I've missed some documentation, this
 is probably a nice little project for the post-4.2 era  :)

Re: www.gnuplot.info

[Clark G. <ga...@di...>]

On the "is shell down why rsync down" question: yes, same service.

On Wed, 21 Feb 2007 00:24:04 +0100 (CET), "Petr Mikulik"
<mi...@ph...> said:
> > By the way.  I made a quick pass over the FAQ and committed the
> > revision to CVS.  But I don't actually know the procedure for 
> > propagating the revised *.tex document to other media. Once the
> > shell server is back up, it will be possible to upload a new
> > html file.  But what about pure text and pdf versions - is there
> > a list of places they need to go?
> 
> Into gnuplot cvs: FAQ (the text)
> Into web page: see gnuplot.sf.net/faq/

Could we update the web pages throughout the hierarchy so that the URLs
that have http://gnuplot.sourceforge.net are relative instead of
absolute? When the web page was at the end of a DSL line, this made
sense, but now that we are back in business we should expunge those
absolute URLs I think.

> Into ftp.gnuplot.info: there is a faq/ directory, I propose to remove
>    it, or write "go to web"

This raises the question: do we really still need ftp access? I like the
legacy files (notably the gnuplot-historical), but we could also move
those to the web repository. I haven't checked about disk space, so
maybe we shouldn't do that, but I wonder whether we really still need
ftp access as a protocol.

--ckg
--
Clark Gaylord
Blacksburg, VA USA
ga...@di...

Re: it seems my Gnuplot 4.0 didn't read my ~/.Xdefaults

[Daniel J S. <dan...@ie...>]

Hans pointed me to the "man X" documentation for details about Xdefaults.  Since that documentation says:

$HOME/.Xdefaults-<hostname>

it might be good that gnuplot's documentation state the same, rather than just .Xdefaults.  (Or, not mention the file at all and instead give the "xrdb" utility.)

Dan

Re: it seems my Gnuplot 4.0 didn't read my ~/.Xdefaults

[Daniel J S. <dan...@ie...>]

Ethan Merritt wrote:

> Because the default for color displays in not to use dashed lines.
> To force it on:
> 
>    echo "gnuplot*dashed: on" | xrdb -merge

Oh, that might do it.  I'll pass that along.  Thanks.

>>I was wondering if some X resource of his has those lines set all solid
> 
> 
> That's possible also, but then you wouldn't get dashed lines in mono either.

Good point.

Dan

Re: it seems my Gnuplot 4.0 didn't read my ~/.Xdefaults

[Ethan M. <merritt@u.washington.edu>]

On Wednesday 21 February 2007 14:05, Ethan Merritt wrote:
> 
> Because the default for color displays in not to use dashed lines.
> To force it on:
> 
>    echo "gnuplot*dashed: on" | xrdb -merge

To expand a bit - the above command will change the default to be
dashed.  Whether or not you change the default, you can still toggle
it inside gnuplot by saying
	set term x11 {dashed|solid}

To make sure you have a set of dash patterns loaded:

	xrdb -merge < /home/local/lib/X11/app-defaults/Gnuplot.app-defaults

(or wherever you put them)

-- 
Ethan A Merritt                                 Courier Deliveries: 1959 NE Pacific
Dept of Biochemistry    M/S 357742
Health Sciences Building
University of Washington - Seattle

Re: it seems my Gnuplot 4.0 didn't read my ~/.Xdefaults

[Ethan M. <merritt@u.washington.edu>]

On Wednesday 21 February 2007 13:50, Daniel J Sebald wrote:
> Around 2/24/06 is a discussion with the subject as I copied to this email header.
> Has this been resolved, if only it means a change to x11.trm documentation?

What is there to resolve, exactly?

> I'm trying to help John Eaton figure out why he doesn't get dashed color lines
> using "set term x11 dashed".  

Because the default for color displays in not to use dashed lines.
To force it on:

   echo "gnuplot*dashed: on" | xrdb -merge


> I was wondering if some X resource of his has those lines set all solid

That's possible also, but then you wouldn't get dashed lines in mono either.


-- 
Ethan A Merritt                                 Courier Deliveries: 1959 NE Pacific
Dept of Biochemistry    M/S 357742
Health Sciences Building
University of Washington - Seattle

Re: it seems my Gnuplot 4.0 didn't read my ~/.Xdefaults

[Daniel J S. <dan...@ie...>]

Around 2/24/06 is a discussion with the subject as I copied to this email header.  Has this been resolved, if only it means a change to x11.trm documentation?  I'm trying to help John Eaton figure out why he doesn't get dashed color lines using "set term x11 dashed".  He gets dashed lines for monochrome.  I was wondering if some X resource of his has those lines set all solid, or dashed is turned off, or something.  But I don't know where to point him to.

My own .Xdefaults didn't do anything so I printed out file names in gplt_x11.c and found .Xdefaults-local is what gplt_x11.c is looking for.  Changing my X defaults from .Xdefaults to .Xdefaults-local fixes the problem.  I searched previous discussions for "Xdefaults-" and found the 2/24/06 thread.

The discussion reveals quite a few options, but the x11.trm documentation basically says .Xdefaults which helps only a little in this case.  Maybe it should be worded ".Xdefaults or any of its variants".  Or will that only send people on an even worse quest?

OK, use color.  :-)  But there is still the issue of other things that one might want to put in .Xdefaults but it has no effect if gnuplot_x11 is looking for .Xdefaults-local and the user doesn't know that.  (I can't seem to find any unix documentation on the Internet about filename variations for .Xdefaults.)

Dan

Re: Is it a bug? Label problem when using big size (mp terminal)

[Daniel J S. <dan...@ie...>]

Daniel J Sebald wrote:

> I seem to recall discussion that one isn't supposed to use size
> greater than 1.  If I'm mistaken, then yes this is a bug with
> clipping text to the range of 1,1 on the plot.

x11 seems to have even worse problems:

gnuplot
plot "example.dat" using ($1):($2) notitle
set size 2, 2.5
replot

Dan

Re: Is it a bug? Label problem when using big size (mp terminal)

[Daniel J S. <dan...@ie...>]

Pen...@gm... wrote:
> Hi,
> 
> I have the following gnuplot script and data file below this email.
> 
> I compiled it with the following command.
> $ gnuplot example.gp
> $ mpost -tex=latex example.mp
> $ mptopdf example.0
> 
> The label on the x axis only shows from -70 to -30, which is smaller
> than the data range. The label shown on the y axis does not cover all
> the data range either. Is it a bug?

What I'm seeing is a correct plot except that the tic annotation stops at what is probably size 1,1.  (Rotation of tic text has nothing to do with the issue.)

I seem to recall discussion that one isn't supposed to use size greater than 1.  If I'm mistaken, then yes this is a bug with clipping text to the range of 1,1 on the plot.

Dan


> 
> Thanks,
> Peng
> 
> $ cat example.gp
> set terminal mp monochrome latex
> set output "example.mp"
> set size 2, 2.5
> set xtics rotate
> plot "example.dat" using ($1):($2) notitle
> 
> $ cat example.dat
> -70 -3
> 20 4

Re: libgd-2.0.34 released

[Ethan M. <merritt@u.washington.edu>]

On Wednesday 21 February 2007 10:18, Don Taber wrote:
> New gd headers no longer #define S_ENC_UTF8 
> and thus break gd.trm.

Thanks for the bug report.
In fact the S_ENC_UTF8 test was not expected to resolve
against gd.h, but against a revised version of gnuplot's
own src/tables.h that should have been part of the same
patch. Unfortunately my patchsets got out of sync.
It's good that you caught it in time!

	Ethan

-- 
Ethan A Merritt                                 Courier Deliveries: 1959 NE Pacific
Dept of Biochemistry    M/S 357742
Health Sciences Building
University of Washington - Seattle

Re: libgd-2.0.34 released

[Don T. <dt...@to...>]

New gd headers no longer #define S_ENC_UTF8 
and thus break gd.trm.  The following patch
lets it compile, but I am not sure it won't
break something. Probably someone who knows
about gd needs to figure out the real fix.

Don Taber

--- gd.trm.~1.110.~     2007-02-09 04:00:48.000000000 -0800
+++ gd.trm      2007-02-21 10:04:56.000000000 -0800
@@ -1876,7 +1876,8 @@
 #ifdef gdFTEX_Adobe_Custom
            /* libgd defaults to UTF-8 encodings. We have limited options for   */
            /* over-riding this, but we can try                                 */
-           if (encoding != S_ENC_UTF8 && ENHgd_font && !strcmp(ENHgd_font,"Symbol")) {
+           //if (encoding != S_ENC_UTF8 && ENHgd_font && !strcmp(ENHgd_font,"Symbol")) {
+           if (ENHgd_font && !strcmp(ENHgd_font,"Symbol")) {
                PNG_FONT_INFO.flags |= gdFTEX_CHARMAP;
                PNG_FONT_INFO.charmap = gdFTEX_Adobe_Custom;
                /* fprintf(stderr,"PNG_set_font: requesting a special charmap\n"); */

Re: www.gnuplot.info

[Petr M. <mi...@ph...>]

> By the way.  I made a quick pass over the FAQ and committed the
> revision to CVS.  But I don't actually know the procedure for 
> propagating the revised *.tex document to other media. Once the
> shell server is back up, it will be possible to upload a new
> html file.  But what about pure text and pdf versions - is there
> a list of places they need to go?

Into gnuplot cvs: FAQ (the text)
Into web page: see gnuplot.sf.net/faq/
Into ftp.gnuplot.info: there is a faq/ directory, I propose to remove
   it, or write "go to web"
Into binary distributions: e.g. in Win there is docs/FAQ.{html,txt}

---
PM

Is it a bug? Label problem when using big size (mp terminal)

[<Pen...@gm...>]

Hi,

I have the following gnuplot script and data file below this email.

I compiled it with the following command.
$ gnuplot example.gp
$ mpost -tex=latex example.mp
$ mptopdf example.0

The label on the x axis only shows from -70 to -30, which is smaller
than the data range. The label shown on the y axis does not cover all
the data range either. Is it a bug?

Thanks,
Peng

$ cat example.gp
set terminal mp monochrome latex
set output "example.mp"
set size 2, 2.5
set xtics rotate
plot "example.dat" using ($1):($2) notitle

$ cat example.dat
-70 -3
20 4