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

Re: pause mouse does not work on windows term on gnuplot.exe (on wgnuplot, pause mouse works correct)

[Tatsuro M. <tma...@ya...>]

Hello

I have tried some modification on mouse.c but got no success.
However, I noticed that pressing keyboard successfully breaks 'pause mouse'. 

Regards

Tatsuro

--- Tatsuro MATSUOKA wrote:

> Hello
> 
> In the octave mailing list, 
> it is revealed that pause mouse does not work correct on gnuplot.exe.
> 
> http://old.nabble.com/Problem-with-ginput-to26706688.html
> 
> Ben Abbott reduce the problem of gnuplot as,
> 
> gnuplot> set term windows 
> gnuplot> plot sin(x) 
> gnuplot> pause mouse 
> gnuplot> print MOUSE_X         
> gnuplot> print MOUSE_Y 
> 
> On the wgnuplot.exe (4.4 and 4.5), the above worked correct as,
> gnuplot> set term windows
> Terminal type set to 'windows'
> Options are 'color noenhanced font "Arial, 10"'
> gnuplot> plot sin(x)
> gnuplot> pause mouse
> paused
> gnuplot>
> gnuplot> print MOUSE_X
> -5.02091446704168
> gnuplot> print MOUSE_Y
> -0.362648407563289
> gnuplot>
> 
> On gnuplot.exe wxt term, it also work 
> 
> Terminal type set to 'wxt'
> gnuplot> plot sin(x)
> gnuplot> pause mouse
> gnuplot>
> gnuplot> print MOUSE_X
> -1.10333064009337
> gnuplot> print MOUSE_Y
> 0.210836991070077
> gnuplot>
> 
> However on windows term on gnuplot.exe (4.4 and 4.5),
> plot sin(x)
> pause mouse
> 
> No response even clicking mouse on the plot window.
> 
> I have paste the commands to the gnuplot prompt using quick edit mode of cmd facilities.
> gnuplot> plot sin(x)
> gnuplot> pause mouse
> gnuplot> print MOUSE_X
>          undefined variable: MOUSE_X
> 
> gnuplot> print MOUSE_Y
>          undefined variable: MOUSE_Y
> 
> 'pause mouse' was skipped.
> 
> At the moment, I have not seen the code concerning to 'pause mouse'.
> 
> Regards
> 
> Tatsuro
> 
> 
> 
>                                  
> 
> --------------------------------------
> Get Disney character's mail address on Yahoo! Mail
> http://pr.mail.yahoo.co.jp/disney/
> 
> ------------------------------------------------------------------------------
> Return on Information:
> Google Enterprise Search pays you back
> Get the facts.
> http://p.sf.net/sfu/google-dev2dev
> _______________________________________________
> gnuplot-beta mailing list
> gnu...@li...
> https://lists.sourceforge.net/lists/listinfo/gnuplot-beta
> 


--------------------------------------
Get Disney character's mail address on Yahoo! Mail
http://pr.mail.yahoo.co.jp/disney/

How to compile gnuplot with AquaTerm?

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

Hello,

on my previous computer (OS X Leopard) the AquaTerm has been built and
used by default. After changing the computer the AquaTerm doesn't get
built into gnuplot any more.

The ./configure reports
  aqua terminal (MacOS X): yes
but the terminal is not available after I make the executable.

Any hints about this? I can send some log, but it might make sense if
someone can suggest what logs to send or what to look for.

Thanks a lot,
   Mojca

pause mouse does not work on windows term on gnuplot.exe (on wgnuplot, pause mouse works correct)

[Tatsuro M. <tma...@ya...>]

Hello

In the octave mailing list, 
it is revealed that pause mouse does not work correct on gnuplot.exe.

http://old.nabble.com/Problem-with-ginput-to26706688.html

Ben Abbott reduce the problem of gnuplot as,

gnuplot> set term windows 
gnuplot> plot sin(x) 
gnuplot> pause mouse 
gnuplot> print MOUSE_X         
gnuplot> print MOUSE_Y 

On the wgnuplot.exe (4.4 and 4.5), the above worked correct as,
gnuplot> set term windows
Terminal type set to 'windows'
Options are 'color noenhanced font "Arial, 10"'
gnuplot> plot sin(x)
gnuplot> pause mouse
paused
gnuplot>
gnuplot> print MOUSE_X
-5.02091446704168
gnuplot> print MOUSE_Y
-0.362648407563289
gnuplot>

On gnuplot.exe wxt term, it also work 

Terminal type set to 'wxt'
gnuplot> plot sin(x)
gnuplot> pause mouse
gnuplot>
gnuplot> print MOUSE_X
-1.10333064009337
gnuplot> print MOUSE_Y
0.210836991070077
gnuplot>

However on windows term on gnuplot.exe (4.4 and 4.5),
plot sin(x)
pause mouse

No response even clicking mouse on the plot window.

I have paste the commands to the gnuplot prompt using quick edit mode of cmd facilities.
gnuplot> plot sin(x)
gnuplot> pause mouse
gnuplot> print MOUSE_X
         undefined variable: MOUSE_X

gnuplot> print MOUSE_Y
         undefined variable: MOUSE_Y

'pause mouse' was skipped.

At the moment, I have not seen the code concerning to 'pause mouse'.

Regards

Tatsuro



                                 

--------------------------------------
Get Disney character's mail address on Yahoo! Mail
http://pr.mail.yahoo.co.jp/disney/

Re: bug? in emf enhanced terminal; 'ylabel' on 4.4rc1 and 4.5

[Tatsuro M. <tma...@ya...>]

Hello

I have built binaries from the latest cvs source 

2009-12-08  Ethan A Merritt  <merritt@u.washington.edu>

	* term/emf.trm (ENHemf_put_text): Left/right/center justification of
	enhanced text was not taking into account the text angle. Now it does.
	Bugfix.

	* src/breaders.c (gd_filetype_function): Make load of jpeg or gif image
	conditional on support by the installed libgd.

The test

set term emf enh
set ylabel 'y label a^{-1}'
set out 'ylabel.emf'
plot sin(x)
set out

worked fine!

Thanks!

The binaries on Cygwin, MinGW and DJGPP have been updated.
http://www.tatsuromatsuoka.com/gnuplot/Eng/cygbin/
http://www.tatsuromatsuoka.com/gnuplot/Eng/winbin/


Regards

Tatsuro


--- Tatsuro MATSUOKA wrote:

> Hello
> 
> 
> Thank you for your immediate response.
> I have apply your patch to the cvs source and tested on MinGW.
> 
> Your patch makes the test goes well.
> 
> Thanks!!
> 
> 
> Regards
> 
> Tatsuro
> 
>  


--------------------------------------
Get Disney character's mail address on Yahoo! Mail
http://pr.mail.yahoo.co.jp/disney/

Re: gnuplot 4.4rc-1 : ReadmeWindows lacks explanation of 'gnuplot.exe'

[Ethan M. (sfeam) <eam...@gm...>]

On Monday 30 November 2009, Petr Mikulik wrote:
> > BTW, I have a quick look it and found that the document 'ReadmeWindows' 
> > lacks explanation of 'gnuplot.exe'.  This is one of the important topic of 
> > gnuplot 4.4.
> 
> Thanks for spotting this. I propose the following text:
> 
> gnuplot.exe: Text (console) mode version of the gnuplot executable with all
> the associated pipe functionality as it is usual on all other platforms.
> This means that this program can also accept commands on stdin (standard
> input) and print messages on stdout (standard output). This program replaces
> pgnuplot.exe and it is recommended to use it for 3rd party applications
> using gnuplot as graph engine, for example Octave (www.octave.org).
> 
> Feel free to comment/update.

Please go ahead and add this, along with any other description you think
would be useful.

	Ethan

Re: bug? in emf enhanced terminal; 'ylabel' on 4.4rc1 and 4.5

[Tatsuro M. <tma...@ya...>]

Hello


Thank you for your immediate response.
I have apply your patch to the cvs source and tested on MinGW.

Your patch makes the test goes well.

Thanks!!


Regards

Tatsuro

 

--- Ethan Merritt  wrote:

> On Monday 07 December 2009 20:56:28 Tatsuro MATSUOKA wrote:
> > Hello
> > 
> > I have tested the following script
> > 
> > set term emf enh
> > set ylabel 'y label a^{-1}'
> > set out 'ylabel.emf'
> > plot sin(x)
> > set out
> > 
> > gnuplot> set term emf enh
> > Terminal type set to 'emf'
> > Options are 'color dashed "Arial" 12 enhanced '
> > gnuplot> set ylabel 'y label a^{-1}'
> > gnuplot> set out 'ylabel.emf'
> > gnuplot> plot sin(x)
> >          warning: emf_move: (-360,10421) out of range
> > gnuplot> set out
> > 
> > 
> >          warning: emf_move: (-360,10421) out of range
> 
> I confirm this.  
> It is a general problem with emf, not specific to windows.
> 
> The enhanced text mode is ignoring text rotation when it does
> left/right/center justification.  The text placement is
> always incorrect when the text is both rotated and enhanced.
> 
> The attached patch is better than the current code, but it is only 
> an approximation to the correct placement.  I do not know how to
> do a more complete treatment of rotation + justification in emf.
> 
> 	Ethan
> 
> 
> 
> > result graph is 
> > http://www.geocities.jp/tmgpltwin/Files/Files.html#0040
> > 0040 ylabel_emf_enh.png
> > 
> > The above confirmed 4.5 on MinGW  and cygwin (changelog 2009-12-05) and gnuplot 4.4rc1(windows
> by
> > Petr) .
> > 
> > 
> > If I do not use a^{-1}, ylabel is located in proper position. 
> > 
> > Regards
> > 
> > Tatsuro
> > --- gnuplot/term/emf.trm	2009-09-17 08:51:22.000000000 -0700
> +++ gnuplot-cvs/term/emf.trm	2009-12-08 10:47:53.000000000 -0800
> @@ -1583,15 +1583,23 @@ ENHemf_put_text(unsigned int x, unsigned
>  	ENHemf_font = emf_fontname;
>  	ENHemf_fontsize = emf_fontsize;
>  
> +    /* EAM -FIXME
> +     * To do proper justification requires some way to track the current position
> +     * after writing a text fragment.  I don't know how to do that in emf.
> +     * So I punt by just estimating the total length.
> +     */
>      if (emf_justify == RIGHT) {
> -	EMF_MoveToEx(x - term->h_char * len, term->ymax-y);
> -	EMF_move(x - EMF_AVG_WID * term->h_char * len, y);
> +	EMF_move(
> +		x - term->h_char * len * cos(emf_vert_text * EMF_10THDEG2RAD),
> +		y - term->h_char * len * sin(emf_vert_text * EMF_10THDEG2RAD)
> +		);
>      } else if (emf_justify == CENTRE) {
> -	EMF_MoveToEx(x - term->h_char * len/2, term->ymax-y);
> -	EMF_move(x - EMF_AVG_WID * term->h_char * len/2, y);
> +	EMF_move(
> +		x - term->h_char * len/2 * cos(emf_vert_text * EMF_10THDEG2RAD),
> +		y - term->h_char * len/2 * sin(emf_vert_text * EMF_10THDEG2RAD)
> +		);
>      } else {
> -	EMF_MoveToEx(x, term->ymax-y);
> -	EMF_move(x, y);
> +	EMF_move( x, y);
>      }
>  
>      emf_justify = LEFT;
> 


--------------------------------------
Get Disney character's mail address on Yahoo! Mail
http://pr.mail.yahoo.co.jp/disney/

Re: bug? in emf enhanced terminal; 'ylabel' on 4.4rc1 and 4.5

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

On Monday 07 December 2009 20:56:28 Tatsuro MATSUOKA wrote:
> Hello
> 
> I have tested the following script
> 
> set term emf enh
> set ylabel 'y label a^{-1}'
> set out 'ylabel.emf'
> plot sin(x)
> set out
> 
> gnuplot> set term emf enh
> Terminal type set to 'emf'
> Options are 'color dashed "Arial" 12 enhanced '
> gnuplot> set ylabel 'y label a^{-1}'
> gnuplot> set out 'ylabel.emf'
> gnuplot> plot sin(x)
>          warning: emf_move: (-360,10421) out of range
> gnuplot> set out
> 
> 
>          warning: emf_move: (-360,10421) out of range

I confirm this.  
It is a general problem with emf, not specific to windows.

The enhanced text mode is ignoring text rotation when it does
left/right/center justification.  The text placement is
always incorrect when the text is both rotated and enhanced.

The attached patch is better than the current code, but it is only 
an approximation to the correct placement.  I do not know how to
do a more complete treatment of rotation + justification in emf.

	Ethan



> result graph is 
> http://www.geocities.jp/tmgpltwin/Files/Files.html#0040
> 0040 ylabel_emf_enh.png
> 
> The above confirmed 4.5 on MinGW  and cygwin (changelog 2009-12-05) and gnuplot 4.4rc1(windows by
> Petr) .
> 
> 
> If I do not use a^{-1}, ylabel is located in proper position. 
> 
> Regards
> 
> Tatsuro

bug? in emf enhanced terminal; 'ylabel' on 4.4rc1 and 4.5

[Tatsuro M. <tma...@ya...>]

Hello

I have tested the following script

set term emf enh
set ylabel 'y label a^{-1}'
set out 'ylabel.emf'
plot sin(x)
set out

gnuplot> set term emf enh
Terminal type set to 'emf'
Options are 'color dashed "Arial" 12 enhanced '
gnuplot> set ylabel 'y label a^{-1}'
gnuplot> set out 'ylabel.emf'
gnuplot> plot sin(x)
         warning: emf_move: (-360,10421) out of range
gnuplot> set out


         warning: emf_move: (-360,10421) out of range

result graph is 
http://www.geocities.jp/tmgpltwin/Files/Files.html#0040
0040 ylabel_emf_enh.png

The above confirmed 4.5 on MinGW  and cygwin (changelog 2009-12-05) and gnuplot 4.4rc1(windows by
Petr) .


If I do not use a^{-1}, ylabel is located in proper position. 

Regards

Tatsuro

--------------------------------------
Get Disney character's mail address on Yahoo! Mail
http://pr.mail.yahoo.co.jp/disney/

Re: gnuplot 4.4 rc1 build failure

[Ethan M. (sfeam) <eam...@gm...>]

On Sunday 06 December 2009, mw...@us... wrote:
> Hi All,
> 
> I tried to build 4.4 RC1 on Solaris 9.  Here are the libraries I have.
> 
> zlib 1.2.3
> libpng 1.2.18
> freetype 2.3.5
> gd 2.0.33
> gcc 3.4.6 or gcc 4.2.?  ( I will try with Sun compiler soon)
> 
> The failure came from the file breader.c.  I got a link err for a call to GDImageCreateFromJPEG.   I don't have JPEG libraries as part of the GD library.  It looks like some pre-processor check need to be added around this call and maybe others? 

I don't see any other places outside of gd.trm itself.
Please try the attached patch.

> config.h has the proper settings for HAVE_GD_PNG and HAVE_GD_JPEG is undefined.  So the configure script picked-up on the fact the JPEG libraries were not available.
> 
> The ran quick tests after commenting out the 1 offending call.  The X11 and PNG seem to work.  I didn't test any other terminals yet.  I do get a warning from 'set term png' about the 'arial' font not being found.  I don't think you could find one font that is available on every operating system.

Right, although the very latest libgd (2.0.37) itself does a more thorough job of querying the system for font information via fontconfig.

	Ethan  

gnuplot 4.4 rc1 build failure

[<mw...@us...>]

Hi All,

I tried to build 4.4 RC1 on Solaris 9.  Here are the libraries I have.

zlib 1.2.3
libpng 1.2.18
freetype 2.3.5
gd 2.0.33
gcc 3.4.6 or gcc 4.2.?  ( I will try with Sun compiler soon)

The failure came from the file breader.c.  I got a link err for a call to GDImageCreateFromJPEG.   I don't have JPEG libraries as part of the GD library.  It looks like some pre-processor check need to be added around this call and maybe others? 

config.h has the proper settings for HAVE_GD_PNG and HAVE_GD_JPEG is undefined.  So the configure script picked-up on the fact the JPEG libraries were not available.

The ran quick tests after commenting out the 1 offending call.  The X11 and PNG seem to work.  I didn't test any other terminals yet.  I do get a warning from 'set term png' about the 'arial' font not being found.  I don't think you could find one font that is available on every operating system.

Mike

Re: gridding: differences of 4.2 and 4.4

[Philipp K. J. <ja...@ie...>]

It was me who wrote the code for the "new" dgrid3d 
function, a couple of years back. I am just looking at
the code and trying to get back into what I tried to do.

>
> Could someone please clarify:
>
> - The BNF description of syntax under "help set dgrid3d" shows
>   only the "hann" option taking 2 additional parameters.
>   But the text says that 5 options use these parameters.
>   Which is correct?

All of the "new" smoothing kernels take the same number
of parameters. There is a set of square brackets missing
in the doc. (My bad.)

>
> - Is there a difference between
> 	set dgrid3d <rows>, <cols>, <norm>        # old syntax
>   and
> 	set dgrid3d <rows>, <cols>, <norm> qnorm  # is this legal?

This should not be legal. (But it might work - got to try.)

>   and
> 	set dgrid3d <rows>, <cols> qnorm <norm>   # new syntax

This should be identical to the first.

>
> - Note that the current BNF description of syntax would not allow
>   either of the first two commands above

That is correct. The synopsis shows only the new form
of the command, which makes the kernel explicit. 
According to the 2nd to last para in the doc, the first 
version is accepted for backward compatibility

I'll look at Ethan's patch next.

>
> 	Ethan

gnuplot 4.4.0 rc-1 binaries on cygwin-1.7, MinGW (native windows) and DJGPP are uploaded on the web

[Tatsuro M. <tma...@ya...>]

Hello

I have uploaded gnuplot 4.4.0 rc-1 binaries on cygwin-1.7, MinGW (native winodws) and DJGPP on my web
site for  testing purpose.

 1. pngcairo/pdfcairo terminals and lua/tikz terminals are supported on Cygwin-1.7 and MinGW
 2. differential package by which wxt terminal are available is prepared for MinGW 
 3. DJGPP binaries (only fundamental terminals but canvas terminal is supported.)

Cygwin
 http://www.tatsuromatsuoka.com/gnuplot/Eng/cygbin/
 gp44rc1-winbinX11-cygwin-1.7.zip

MinGW and DJGPP   
 http://www.tatsuromatsuoka.com/gnuplot/Eng/winbin/
 gp44rc1-winbin.zip, gp44rc1-winbin-wxt-diff.zip, gp44rc1-dj2bin.zip

If you find some problem, please post to gnuplot-beta_atmark_lists.sourceforge.net except for careless
mistake like a packaging error.

Regards

Tatsuro

--------------------------------------
Learn more about HIV/AIDS - Red Ribbon Campaign 2009
http://pr.mail.yahoo.co.jp/redribbon/

Build results of gnuplot 4.4.0 rc-1 with lua patch on cygwin-1.7 and MinGW

[Tatsuro M. <tma...@ya...>]

Hello

I have tried to build gnuplot 4.4.0rc-1 with the patch 'lua_install_30nov2009.patch'.

On the cygwin-1.7, building binaries were done successfully.
However 'make install-strip' did not create /usr/local/share/gnuplot/4.4/lua.
I manually created /usr/local/share/gnuplot/4.4/lua, copy required files there.

The tikz terminal seemed to work correct.


**************
$ patch -p 1 < ../lua_install_30nov2009.patch
patching file src/Makefile.am
patching file term/lua.trm
Hunk #1 succeeded at 43 with fuzz 2.
Hunk #2 succeeded at 542 (offset 13 lines).
Hunk #3 succeeded at 1272 (offset 20 lines).
patching file term/Makefile.am.in
patching file config/makefile.cyg
patching file config/makefile.dj2
patching file config/makefile.mgw
Hunk #1 succeeded at 104 with fuzz 2.
patching file config/makefile.os2
***********
Patch seemed worked correctly.  In the case of cvs source, ./prepare was done.
However there is not prepare script in source of 4.4.
If autotools were used correctly, the 'make install' problem might not occur.

For MinGW, I need to modify makefile.mgw and lua.trm. 
makefile.mgw
 Modify it almost the same as that in 4.5 (The difference is only version number.)
lua.trm
 declaration error in 'len'.
The above are already included in the 4.5. Therefore these are not  problems.

PS 1. pngcairo/pdfcairo terminals worked as expected on both cygwin-1.7 and MinGW. 
     On MinGW, the wxt terminal also worked as expected.

PS 2. I also build DJGPP binary without lua/tikz terminal. It worked correctly.

Regards

Tatsuro



  



 



--------------------------------------
Learn more about HIV/AIDS - Red Ribbon Campaign 2009
http://pr.mail.yahoo.co.jp/redribbon/

Re: patch for makefile.mgw

[Ethan M. (sfeam) <eam...@gm...>]

On Friday 04 December 2009, Tatsuro MATSUOKA wrote:
> Hello
> 
> I found inconsistency in makefile.mgw in the current cvs source.
> In install section of makefile.mgw
> 
> 	mkdir -p $(DESTDIR)/share/PostScript
> 	cp ../term/PostScript/*.ps $(DESTDIR)/$(GNUPLOT_PS_DIR)
> 
> The above does not work correct if  GNUPLOT_PS_DIR is not share/PostScipt.
> 
> Is the above to be the following
> 	mkdir -p $(DESTDIR)/$(GNUPLOT_PS_DIR)
> 	cp ../term/PostScript/*.ps $(DESTDIR)/$(GNUPLOT_PS_DIR)
> 
> 
> I made a patch for makefile.mgw according the above.

OK. Applied to 4.4 and 4.5

> Perhaps similar correction should be made for makefile.cyg.

Yes. That looks correct. Actually it seems there is an error in the
order of directory creation in makefile.cyg, but this will fix it.

	Ethan


> 
> Regards
> 
> Tatsuro

Re: gridding: differences of 4.2 and 4.4

[Ethan M. (sfeam) <eam...@gm...>]

On Friday 04 December 2009, Ethan Merritt wrote:
> On Friday 04 December 2009 08:19:43 Petr Mikulik wrote:
>  
> > There is yet another error in 4.4 -- the following command fails:
> > m=5; n=8; set dgrid3d m,m,n
> 
> I'll have a quick go at fixing this.

Patch attached.
It turned out to be more involved than I expected, so please test.

	Ethan

Re: binary.dem triggers error in steps.dem ? on the cvs source (2009-12-03 )

[Tatsuro M. <tma...@ya...>]

Hello

--- Ethan Merritt  wrote:

> On Friday 04 December 2009 08:07:00 Ethan Merritt wrote:
> > On Friday 04 December 2009, Tatsuro MATSUOKA wrote:
> > > Hello
> > > I am trying to build gnuplot binary from the on the cvs source (2009-12-03 ).
> > > 
> > > 'make check' on cygwin and load 'all.dem' on (windows and wxt terminals) MinGW failed at
> > > 
> > > ******************** file steps.dem ********************
> > > "steps.dem", line 14: this 2D plot style cannot handle matrix data
> > 
> > This must be a side effect of the checks for matrix data that I added yesterday.
> > I suspect that the "df_matrix" flag is not being re-initialized properly at
> > the start of each plot.
> 
> I think I have fixed this.
> There was indeed an un-initialized variable (df_num_bin_records).
> It was being set by the "binary" demo, but never reset to 0 afterward.
> Please retry with current CVS.

I have noticed your fix. The current cvs works fine on Cygwin, MinGW and DJGPP.  Before I will reply
to the previous post, I have just found your maii :-).  

New binaries have been just uploaded  on my web.

I appreciate for your efforts.

BTW, now I am trying to build gnuplot-4.4-rc1 with your patch for lua terminal for Cygwin-1.7 and
MinGW.

I will report results after I will finished the trials.

Regards

Tatsuro  

--------------------------------------
Learn more about HIV/AIDS - Red Ribbon Campaign 2009
http://pr.mail.yahoo.co.jp/redribbon/

Re: binary.dem triggers error in steps.dem ? on the cvs source (2009-12-03 )

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

On Friday 04 December 2009 08:07:00 Ethan Merritt wrote:
> On Friday 04 December 2009, Tatsuro MATSUOKA wrote:
> > Hello
> > I am trying to build gnuplot binary from the on the cvs source (2009-12-03 ).
> > 
> > 'make check' on cygwin and load 'all.dem' on (windows and wxt terminals) MinGW failed at
> > 
> > ******************** file steps.dem ********************
> > "steps.dem", line 14: this 2D plot style cannot handle matrix data
> 
> This must be a side effect of the checks for matrix data that I added yesterday.
> I suspect that the "df_matrix" flag is not being re-initialized properly at
> the start of each plot.

I think I have fixed this.
There was indeed an un-initialized variable (df_num_bin_records).
It was being set by the "binary" demo, but never reset to 0 afterward.
Please retry with current CVS.

	Ethan

Re: gridding: differences of 4.2 and 4.4

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

On Friday 04 December 2009 13:25:18 Ethan Merritt wrote:
> On Friday 04 December 2009 08:19:43 Petr Mikulik wrote:

>  
> > There is yet another error in 4.4 -- the following command fails:
> > m=5; n=8; set dgrid3d m,m,n
> 
> I confirm this one.
> It looks like the "new" set_dgrid3d() routine is falsely assuming
> it can detect numerical values by calling isanumber().
> 
> <aside>
>  We should rename that routine!
>  It doesn't at all do what its name suggests.
> </aside>
> 
> I'll have a quick go at fixing this.

Hmm. This is more complicated than I thought.

Could someone please clarify:

- The BNF description of syntax under "help set dgrid3d" shows
  only the "hann" option taking 2 additional parameters.
  But the text says that 5 options use these parameters.
  Which is correct?

- Is there a difference between
	set dgrid3d <rows>, <cols>, <norm>        # old syntax
  and
	set dgrid3d <rows>, <cols>, <norm> qnorm  # is this legal?
  and
	set dgrid3d <rows>, <cols> qnorm <norm>   # new syntax

- Note that the current BNF description of syntax would not allow
  either of the first two commands above

	Ethan

Re: gridding: differences of 4.2 and 4.4

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

On Friday 04 December 2009 08:19:43 Petr Mikulik wrote:
> An anonymous user "r/e-letter/inpost" reported different outputs of gnuplot 
> from OpenSUSE and 4.4. Well it seems that the 3rd parameter of the gridding 
> method changed between 4.2 and 4.4.
> 
> Try any 4.2.x and 4.4 and these examples:
> 
> set dgrid3d ,,1; show dgrid; splot '3.dat' with line
> set dgrid3d ,,8; show dgrid; splot '3.dat' with line

I get identical results with 4.2.5 and 4.4.0-rc1.

Perhaps you could post screenshots?
 
> There is yet another error in 4.4 -- the following command fails:
> m=5; n=8; set dgrid3d m,m,n

I confirm this one.
It looks like the "new" set_dgrid3d() routine is falsely assuming
it can detect numerical values by calling isanumber().

<aside>
 We should rename that routine!
 It doesn't at all do what its name suggests.
</aside>

I'll have a quick go at fixing this.


	Ethan

gridding: differences of 4.2 and 4.4

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

An anonymous user "r/e-letter/inpost" reported different outputs of gnuplot 
from OpenSUSE and 4.4. Well it seems that the 3rd parameter of the gridding 
method changed between 4.2 and 4.4.

Try any 4.2.x and 4.4 and these examples:

set dgrid3d ,,1; show dgrid; splot '3.dat' with line
set dgrid3d ,,8; show dgrid; splot '3.dat' with line

3.dat:
1 1 1
1 2 2
1 3 3

2 1 2
2 2 2
2 3 2

3 1 3
3 2 2
3 3 1


There is yet another error in 4.4 -- the following command fails:
m=5; n=8; set dgrid3d m,m,n

---
PM

Re: binary.dem triggers error in steps.dem ? on the cvs source (2009-12-03 )

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

On Friday 04 December 2009, Tatsuro MATSUOKA wrote:
> Hello
> I am trying to build gnuplot binary from the on the cvs source (2009-12-03 ).
> 
> 'make check' on cygwin and load 'all.dem' on (windows and wxt terminals) MinGW failed at
> 
> ******************** file steps.dem ********************
> "steps.dem", line 14: this 2D plot style cannot handle matrix data

This must be a side effect of the checks for matrix data that I added yesterday.
I suspect that the "df_matrix" flag is not being re-initialized properly at
the start of each plot.
Thank you for finding this.

	Ethan




> Interestingly I execute load 'steps.dem' as first command on gnuplot, it worked correctly.
> 
> In all.dem, preceeding demo to 'steps.dem' is 'binary.dem'.
> 
> So I tried
> 
> gnuplot> pwd
> /cygdrive/d/usr/Tatsu/cyghome-1.7/gnuplotcvs/demo
> gnuplot> load 'steps.dem'
> Hit return for demonstration of automatic histogram creation
> Hit return for normal distribution function.
> Hit return for inverse error function.
> Hit return for inverse normal distribution function.
> Press return to continue
> gnuplot> load 'binary.dem'
> Hit return to continue (1)
> Hit return to continue (2)
> Hit return to continue (3)
> gnuplot> load 'steps.dem'
>          "steps.dem", line 14: this 2D plot style cannot handle matrix data
> 
> The above is an example on cygwin-1.7.  Similar phenomena occur on MinGW (windows and wxt terminals)
> and DJGPP. 
> 
> As far as I remember this is not occurred by the cvs source  at 2009-12-01.
> 
> I will check using by cvs update -D '2009-12-01'.
> 
> Anyway current source is something wrong at least with three platforms cygwin, MinGW and DJGPP.
> 
> Regards
> 
> Tatsuro
> 
> I
> 
> 
> 
> 
> I cannot understand why such mysterious thing occur.
> 
> Regards
> 
> 
> 
> --------------------------------------
> Learn more about HIV/AIDS - Red Ribbon Campaign 2009
> http://pr.mail.yahoo.co.jp/redribbon/
> 
> ------------------------------------------------------------------------------
> Join us December 9, 2009 for the Red Hat Virtual Experience,
> a free event focused on virtualization and cloud computing. 
> Attend in-depth sessions from your desk. Your couch. Anywhere.
> http://p.sf.net/sfu/redhat-sfdev2dev
> _______________________________________________
> gnuplot-beta mailing list
> gnu...@li...
> https://lists.sourceforge.net/lists/listinfo/gnuplot-beta
> 

Re: binary.dem triggers error in steps.dem ? on the cvs source (2009-12-03 )

[Tatsuro M. <tma...@ya...>]

Hello

I have test by the previous cvs source (2009-12-01)

Terminal type set to 'svga'
gnuplot> cd '../demo'
gnuplot> load 'steps.dem'
Hit return for demonstration of automatic histogram creation
Hit return for normal distribution function.
Hit return for inverse error function.
Hit return for inverse normal distribution function.
Press return to continue
gnuplot> load 'binary.dem'
Hit return to continue (1)
Hit return to continue (2)
Hit return to continue (3)
gnuplot> load 'steps.dem'
Hit return for demonstration of automatic histogram creation
Hit return for normal distribution function.
Hit return for inverse error function.
Hit return for inverse normal distribution function.
Press return to continue

It worked without problem.

Regards

Tatsuro

--- Tatsuro MATSUOKA  wrote:

> Hello
> I am trying to build gnuplot binary from the on the cvs source (2009-12-03 ).
> 
> 'make check' on cygwin and load 'all.dem' on (windows and wxt terminals) MinGW failed at
> 
> ******************** file steps.dem ********************
> "steps.dem", line 14: this 2D plot style cannot handle matrix data
> 
> Interestingly I execute load 'steps.dem' as first command on gnuplot, it worked correctly.
> 
> In all.dem, preceeding demo to 'steps.dem' is 'binary.dem'.
> 
> So I tried
> 
> gnuplot> pwd
> /cygdrive/d/usr/Tatsu/cyghome-1.7/gnuplotcvs/demo
> gnuplot> load 'steps.dem'
> Hit return for demonstration of automatic histogram creation
> Hit return for normal distribution function.
> Hit return for inverse error function.
> Hit return for inverse normal distribution function.
> Press return to continue
> gnuplot> load 'binary.dem'
> Hit return to continue (1)
> Hit return to continue (2)
> Hit return to continue (3)
> gnuplot> load 'steps.dem'
>          "steps.dem", line 14: this 2D plot style cannot handle matrix data
> 
> The above is an example on cygwin-1.7.  Similar phenomena occur on MinGW (windows and wxt
> terminals)
> and DJGPP. 
> 
> As far as I remember this is not occurred by the cvs source  at 2009-12-01.
> 
> I will check using by cvs update -D '2009-12-01'.
> 
> Anyway current source is something wrong at least with three platforms cygwin, MinGW and DJGPP.
> 
> Regards
> 
> Tatsuro
> 
> I
> 
> 
> 
> 
> I cannot understand why such mysterious thing occur.
> 
> Regards
> 
> 
> 
> --------------------------------------
> Learn more about HIV/AIDS - Red Ribbon Campaign 2009
> http://pr.mail.yahoo.co.jp/redribbon/
> 
> ------------------------------------------------------------------------------
> Join us December 9, 2009 for the Red Hat Virtual Experience,
> a free event focused on virtualization and cloud computing. 
> Attend in-depth sessions from your desk. Your couch. Anywhere.
> http://p.sf.net/sfu/redhat-sfdev2dev
> _______________________________________________
> gnuplot-beta mailing list
> gnu...@li...
> https://lists.sourceforge.net/lists/listinfo/gnuplot-beta
> 


--------------------------------------
Learn more about HIV/AIDS - Red Ribbon Campaign 2009
http://pr.mail.yahoo.co.jp/redribbon/

binary.dem triggers error in steps.dem ? on the cvs source (2009-12-03 )

[Tatsuro M. <tma...@ya...>]

Hello
I am trying to build gnuplot binary from the on the cvs source (2009-12-03 ).

'make check' on cygwin and load 'all.dem' on (windows and wxt terminals) MinGW failed at

******************** file steps.dem ********************
"steps.dem", line 14: this 2D plot style cannot handle matrix data

Interestingly I execute load 'steps.dem' as first command on gnuplot, it worked correctly.

In all.dem, preceeding demo to 'steps.dem' is 'binary.dem'.

So I tried

gnuplot> pwd
/cygdrive/d/usr/Tatsu/cyghome-1.7/gnuplotcvs/demo
gnuplot> load 'steps.dem'
Hit return for demonstration of automatic histogram creation
Hit return for normal distribution function.
Hit return for inverse error function.
Hit return for inverse normal distribution function.
Press return to continue
gnuplot> load 'binary.dem'
Hit return to continue (1)
Hit return to continue (2)
Hit return to continue (3)
gnuplot> load 'steps.dem'
         "steps.dem", line 14: this 2D plot style cannot handle matrix data

The above is an example on cygwin-1.7.  Similar phenomena occur on MinGW (windows and wxt terminals)
and DJGPP. 

As far as I remember this is not occurred by the cvs source  at 2009-12-01.

I will check using by cvs update -D '2009-12-01'.

Anyway current source is something wrong at least with three platforms cygwin, MinGW and DJGPP.

Regards

Tatsuro

I




I cannot understand why such mysterious thing occur.

Regards



--------------------------------------
Learn more about HIV/AIDS - Red Ribbon Campaign 2009
http://pr.mail.yahoo.co.jp/redribbon/

patch for makefile.mgw

[Tatsuro M. <tma...@ya...>]

Hello

I found inconsistency in makefile.mgw in the current cvs source.
In install section of makefile.mgw

	mkdir -p $(DESTDIR)/share/PostScript
	cp ../term/PostScript/*.ps $(DESTDIR)/$(GNUPLOT_PS_DIR)

The above does not work correct if  GNUPLOT_PS_DIR is not share/PostScipt.

Is the above to be the following
	mkdir -p $(DESTDIR)/$(GNUPLOT_PS_DIR)
	cp ../term/PostScript/*.ps $(DESTDIR)/$(GNUPLOT_PS_DIR)


I made a patch for makefile.mgw according the above.

Perhaps similar correction should be made for makefile.cyg.

Regards

Tatsuro

--------------------------------------
Learn more about HIV/AIDS - Red Ribbon Campaign 2009
http://pr.mail.yahoo.co.jp/redribbon/

Re: lua.trm for djgpp build

[Tatsuro M. <tma...@ya...>]

Hello Ethan


I understand what you say.

The lua terminal cannot be safely implemented to the DJGPP at present.
According to the guideline that you have shown, one have to implement snprintf by oneself.
It will be possible but I do not want to spend any more time for DJGPP.  

I have not known historical background that you have stated. I have tried to build DJGPP version
gnuplot because you made change in makefile.dj2 for LUA terminal and there existed the problem of
snprintf.  This is an result of my experiment.  Use of sprintf instead of snprintf is not allowable so
that the conclusion  is that the lua/tikz terminals is not officially supported on the DJGPP.   

I am sometimes use DJGPP version of gnuplot as a gnuplot in compact size. It is saved in the USB. 
I only use the emf and eps terminals on DJGPP gnuplot.

Therefore for me, it is OK that DJGPP does not support LUA terminal.

How do you treat the change in makefile.dj2? 

For example

# Similarly for lua scripts used by the lua terminal
GNUPLOT_LUA_DIR = share/lua


I think that it is enough to comment out like 

# Similarly for lua scripts used by the lua terminal
#GNUPLOT_LUA_DIR = share/lua

because it is free that a user will try to build at his own risk.

However, I have no complaint that if you make makefile.dj2 to the previous one.

Regards

Tatsuro

--- Ethan Merritt  wrote:

> On Thursday 03 December 2009 03:03:57 Petr Mikulik wrote:
> > 
> > DJGPP never had snprintf, see config.djg:
> > 
> > /* Define to 1 if you have the `snprintf' function. */
> > /* #undef HAVE_SNPRINTF */
> > 
> > 
> > Thus there should be:
> > 
> > #ifdef HAVE_SNPRINTF
> >     snprintf(...);
> > #else
> >     sprintf(...);
> > #endif
> 
> In my view we are past the point where we can guarantee support for
> platforms too ancient to provide ansi-compliant C environments.
> 
> There is good reason to require snprintf(). Adding #ifdef statements
> to avoid it is an ugly hack that introduces known overflow problems.
> Are you sure that sprintf() itself is ansi-compliant on DJGPP?
> It isn't on ultrix, for instance. That's why the
> "ultrix" patch has been on the contributions page of the
> web site for years now. It allows mousing on ultrix systems even
> though the have a broken sprintf() implementation.
> 
> But really the easiest is to say that DJGPP doesn't support lua/tikz.
> Is anyone set up to test whether a no-snprintf() version actually
> works to produce lua->tikz->pdflatex output correctly on a DJGPP system?
> After all, just because it compiles does not mean that it works correctly.
> 
> 	Ethan
> 
> ------------------------------------------------------------------------------
> Join us December 9, 2009 for the Red Hat Virtual Experience,
> a free event focused on virtualization and cloud computing. 
> Attend in-depth sessions from your desk. Your couch. Anywhere.
> http://p.sf.net/sfu/redhat-sfdev2dev
> _______________________________________________
> gnuplot-beta mailing list
> gnu...@li...
> https://lists.sourceforge.net/lists/listinfo/gnuplot-beta
> 


--------------------------------------
Learn more about HIV/AIDS - Red Ribbon Campaign 2009
http://pr.mail.yahoo.co.jp/redribbon/