gnuplot-bugs Mailing List for gnuplot

„¬„ª„ª„ª„ª„ª„ª„ª„ª„ª„ª„ª„ª„ª„ª„ª„ª„ª„ª„ª„ª„ª„ª„ª„ª„ª„ª„ª„ª„ª„ª„ª„­
 ‹ÆŠENEWS@Vol.032		‹Â“Vƒjƒ…[ƒXƒŒƒ|[ƒ^[(‹­)ƒƒ‹ƒ}
„¯„ª„ª„ª„ª„ª„ª„ª„ª„ª„ª„ª„ª„ª„ª„ª„ª„ª„ª„ª„ª„ª„ª„ª„ª„ª„ª„ª„ª„ª„ª„ª„®
			ƒAƒiƒ^‚ÌŒ³‚ɐM‚¶‚ç‚ê‚È‚¢NEWS‚ð‚¨“Í‚¯I

„¬„Ÿ„­@„¬„Ÿ„­@„¬„Ÿ„­@„¬„Ÿ„­@„¬„Ÿ„­@„¬„Ÿ„­@„¬„Ÿ„­@„¬„Ÿ„­
„ o„ @„ ‰ï„ @„ ‚¢„ @„ Œn„ @„ Ž–„ @„ Œ„ @„ •ë„ @„ I„ 
„¯„Ÿ„®@„¯„Ÿ„®@„¯„Ÿ„®@„¯„Ÿ„®@„¯„Ÿ„®@„¯„Ÿ„®@„¯„Ÿ„®@„¯„Ÿ„®

‰ï‚¦‚éƒTƒCƒg‚ÍŽ–Œ‚ª‹N‚«‚é‰Â”\«‚ª‚ ‚é‚Ì‚¾‚ªA

‚Ü‚Ÿ“–•û‚ÉŒ¾‚킹‚ê‚ΊǗ–â‘è‚Ȃ킯‚ŁB

‘±‚«Fhttp://www.okidokiaiai.com/


@ „¬„ª„ª„ª„ª„ª„ª„ª„ª„ª„ª„ª„ª„ª„ª„ª„ª„ª„ª„ª„ª„ª„ª„ª„ª„ª„ª„ª„ª™œc
„¡š@‹Â“VÌ‹ÆŠENEWSFÅV‹LŽ–ƒsƒbƒNƒAƒbƒv
„¹„ª„ª„ª„ª„ª„ª„ª„ª„ª„ª„ª„ª„ª„ª„ª„ª„ª„ª„ª„ª„ª„ª„ª„ª„ª„ª„ª„ª„ª


 „­@‚Ü‚¸‚̓lƒbƒgŒðÛ‚©‚çƒXƒ^[ƒgu‘åŽèƒ|[ƒ^ƒ‹ƒTƒCƒg‚̉Œ‹‚сv
„¯„®„ª„ª„ª„ª„ª„ª„ª„ª„ª„ª„ª„ª„ª„ª„ª„ª„ª„ª„ª„ª„ª„ª„ª„ª„ª„ª„ª„ª„ª„ª
‘±‚«Fhttp://www.okidokiaiai.com/


 „­@T––‚̃f[ƒg‚Ì‘ŠŽè‚ðƒlƒbƒg‚̏o‰ï‚¢‚Łˏ—«F10l‚É1l
„¯„®„ª„ª„ª„ª„ª„ª„ª„ª„ª„ª„ª„ª„ª„ª„ª„ª„ª„ª„ª„ª„ª„ª„ª„ª„ª„ª„ª„ª„ª„ª
‘±‚«Fhttp://www.okidokiaiai.com/


 „­@“Þ—ÇŒ§‚ª­Žq‰»‘΍ô‚Łuo‰ï‚¢ŒnvH
„¯„®„ª„ª„ª„ª„ª„ª„ª„ª„ª„ª„ª„ª„ª„ª„ª„ª„ª„ª„ª„ª„ª„ª„ª„ª„ª„ª„ª„ª„ª„ª
‘±‚«Fhttp://www.okidokiaiai.com/


„ª„ª„ª„ª„ª„ª„ª„ª„ª„ª„ª„ª„ª„ª„ª„ª„ª„ª„ª„ª„ª„ª„ª„ª„ª„ª„ª„ª„ª„ª„ª„ª„ª„ª„ª„ª„ª„ª
@œ ‹ÆŠENEWS œ
cccccccccccccccccccccccccccccccccccccc
@http://www.okidokiaiai.com/
@‰¤‹­ 

> Igor Bray wrote:
>
> [CC'ing this to -beta, because this is not a bug but a feature proposal=
.]
>
>> modifying the graphics.c file in the way that is listed below. I had
>> hoped that the latest version of gnuplot would give me the required
>> capacity, but I don't believe it does.
>
> Unfortunately, the patch, as attached, is quite unusable --- version
> 3.7.1 is 3 release versions, or 7 years out of date; and the patch
> itself is backwards.
>
> To first approximation, I agree with the reasoning of that patch:
> 'set key spacing <sp>' really should affect the line spacing of the key
> title, too.
>
> Could you please try to re-do this patch, based on the current CVS
> sources, and post it to the patches tracker?

I've written a key patch for enhanced placement that I think is fairly
clean and "decruftifies" the layout a little bit.  Before writing a
conflicting patch, please try that patch and see if we can build off of
it.

Dan

Igor Bray wrote:

[CC'ing this to -beta, because this is not a bug but a feature proposal.]

> modifying the graphics.c file in the way that is listed below. I had
> hoped that the latest version of gnuplot would give me the required
> capacity, but I don't believe it does. 

Unfortunately, the patch, as attached, is quite unusable --- version 
3.7.1 is 3 release versions, or 7 years out of date; and the patch 
itself is backwards.

To first approximation, I agree with the reasoning of that patch:
'set key spacing <sp>' really should affect the line spacing of the key 
title, too.

Could you please try to re-do this patch, based on the current CVS 
sources, and post it to the patches tracker?

Hi, I've been using gnuplot for over a decade and contributed error
bars for polar plots. One issue that has always annoyed me is that I
could not vary the size of the key title box with the "spacing"
command. I often have subscripts and superscripts in the key titles
which get the box lines go through them. Accordingly, I have been
modifying the graphics.c file in the way that is listed below. I had
hoped that the latest version of gnuplot would give me the required
capacity, but I don't believe it does. If others see value in what I
have done perhaps this could be remedied in a future version. An
independent determination of the key title height would be ideal. The
following uses "spacing".

Thanks,

Igor

-- 
---------------------------------------------------------------------
Igor Bray, ARC Australian Professorial Fellow,                _--_|\ 
Physics and Energy Studies,                                  /      \
School of Engineering Science,                               *_.--._/
Murdoch University,             phone(fax): +61 8 9360 6443(6183)  v
90 South Street, Murdoch,       mailto:I....@mu...           
Perth, Western Australia, 6150  http://atom.murdoch.edu.au/
---------------------------------------------------------------------

~/gnuplot-3.7.1> diff graphics.c graphics.c~
*** graphics.c  Tue Feb  5 11:00:51 2002
--- graphics.c~ Wed Sep 15 23:30:29 1999
***************
*** 1050,1054 ****
      /*{{{  calculate the window in the grid for the key */
-       /* Igor Bray modified the code to ensure the key title was the
-          same height as the key entries. This was done by replacing 
-            "t->v_char" with "key_entry_height" in a number of places */
      if (lkey == 1 || (lkey == -1 && key_vpos != TUNDER)) {
--- 1050,1051 ----
***************
*** 1057,1060 ****
        key_w = key_col_wth * key_cols;
!       /*      key_h = (ktitl_lines) * t->v_char + key_rows * key_entry_height;*/
!         key_h = (ktitl_lines) * key_entry_height + key_rows * key_entry_height;
        if (lkey == -1) {
--- 1054,1056 ----
        key_w = key_col_wth * key_cols;
!       key_h = (ktitl_lines) * t->v_char + key_rows * key_entry_height;
        if (lkey == -1) {
***************
*** 1750,1752 ****
            s = ss;
!           yl -= key_entry_height/2; /*t->v_char/2; Igor fixing key title height*/
            while ((e = (char *) strchr(s, '\n')) != NULL) {
--- 1746,1748 ----
            s = ss;
!           yl -= t->v_char / 2;
            while ((e = (char *) strchr(s, '\n')) != NULL) {
***************
*** 1766,1770 ****
                s = ++e;
!               yl -= key_entry_height; /*t->v_char;Igor fixing key title height*/
            }
!           yl += key_entry_height/2; /*t->v_char/2;Igor fixing key title height*/
            free(ss);
--- 1762,1766 ----
                s = ++e;
!               yl -= t->v_char;
            }
!           yl += t->v_char / 2;
            free(ss);
***************
*** 1783,1788 ****
     and first entry                          *//* JFi */
!             /*            (*t->move) (key_xl, key_yt - (ktitl_lines) * t->v_char);    JFi */
!             /*            (*t->vector) (key_xr, key_yt - (ktitl_lines) * t->v_char);  JFi */
!                       (*t->move)  (key_xl,key_yt - (ktitl_lines) * key_entry_height); /* Igor fixing key title height */
!                       (*t->vector)(key_xr,key_yt - (ktitl_lines) * key_entry_height); /* Igor fixing key title height */
        }
--- 1779,1782 ----
     and first entry                          *//* JFi */
!           (*t->move) (key_xl, key_yt - (ktitl_lines) * t->v_char);    /* JFi */
!           (*t->vector) (key_xr, key_yt - (ktitl_lines) * t->v_char);  /* JFi */
        }

Faisal Mathkoor wrote:
> (The c:\gnuplot\bin\wgnuplot.hlp help file was created for a language 
> not supported by your version of windows.)

That was a build error in the very first build of 4.0.0 we released for
Win32.  It's been corrected only days later, but apparently the
corrected version never made it to all the mirrors.

Please get a fresh binary package of gnuplot-4.0.0 for Win32 off the
"Files" section of our sourceforge site and use that.

Dear Sr's,
 
Thank you for gnuplot program. I found a bug whe I use gnuplot on my computer, as 
 
when I try to look up in the help menu, the following message appear to me
 
(The c:\gnuplot\bin\wgnuplot.hlp help file was created for a language not supported by your version of windows.)
 
a fter click on ok the following another message appear 
 
( The instruction at "0x0102a512" referenced memory at "0x0000000c" .The memory could not be "read". click ok to terminate the program 
click cancel to debug the program )
 
after clicking ok the the interface of gnuplot still appear but I can not go into help
 
This bug is for version 4.0 but not for 3.7 which work fine
 
my computer is pentieum III x86 Family 6 model  8 stepping 6 AT/AT compatibale
mu windows is Microsoft windows 2000 
500.2195  Srvice pack 4
 
Faisal Mathkoor










		
---------------------------------
Do you Yahoo!?
 Yahoo! Small Business - Try our new resources site! 

Zack Sullivan wrote:

> will load a binary file of the format:
> <N+1> y0  y1 ...  yN
>  x0  z00 z01 ... z0N
>  x1  z10 z11 ... z1N etc.
> 
> This is not correct.  The actual format appears to be:
> 
> <N>  x0  x1 ...  xN
> y0  z00 z01 ... z0N
> y1  z10 z11 ... z1N etc.

> i.e. a) you should specify that the lines have N z-components, rather 
> than being N+1 single-precision numbers long. 

No. They do have N+1 z-components, thanks to being zero-indexed.

 > b) the x and y axes are swapped vs. the specification.

You appear to be correct on this count, though.  And it's been like that 
for at least 11.5 years now (I've traced it back all the way to version 
3.5 of Sep 1993).   Amazing that nobody noticed this any earlier.  This 
does explain one of the old mysteries of gnuplot, though: why files in 
'binary' mode consistently seemed to get their hidden3d displays 
backside up.

As to which way round to fix this, I'm not quite sure.  The problem is 
that it's hard to know how many people already found out about this 
discrepancy and hardwired the actual behaviour into code of their own.
I guess patching the docs to match reality would be easier on everybody.

I have been using gnuplot 3.7 patchlevel 3 on a linux system recently, but 
I see that this bug still exists in the v. 4.0 documentation and source 
code.

You specify that
> splot "filename" binary
will load a binary file of the format:
<N+1> y0  y1 ...  yN
 x0  z00 z01 ... z0N
 x1  z10 z11 ... z1N etc.

This is not correct.  The actual format appears to be:

<N>  x0  x1 ...  xN
y0  z00 z01 ... z0N
y1  z10 z11 ... z1N etc.

i.e. a) you should specify that the lines have N z-components, rather 
than being N+1 single-precision numbers long. b) the x and y axes are 
swapped vs. the specification.

If you use N+1 instead of N for your 1st entry, the file will not load.

You can confirm the second bug by creating a test file whose data values 
only change with y (or x), and look at the resulting plot.

A temporary fix for people who missed the axis-swapping is to just fix the 
first number to be <N> and use:
splot "filename" binary using ($2):($1):3

Please update the documentation to correct this.  If this is an unexpected 
correction, then I would request that you change the documented 
specification rather than alter the code, as binary data files are 
unpleasant to rewrite.

Aside: Note, that the example binary files are generated by outputting <N> 
as the first number.  Looking at the source code is how I finally 
identified the first bug.  Creating a simple test file is how I identified 
the second.

	Zack Sullivan
	Fermi National Accelerator Laboratory

  Dear gnuplot developers,

encouraged by the manual to report bugs in the gnuplot program, I want to
address today a kind of an inconsistency present in gnuplot version 4.0, but
to my knowledge it exists at least already in version 3.7.
The inconsistency is related to the role of x and y in ordered grid data,
produced by output to the table terminal, best seen when hidden3d option
is set:

1) splotting data files that are ordered like explained in sect. 38.1.2
   "Example Datafile" in the gnuplot 4.0 manual (i.e., ordered in data
   blocks with constant x value and varying y values) works fine: O.K.
   (first plot in my demo file)
2) splotting functions in x and y works fine, O.K. (second plot in demo)
3) splotting functions in x and y to terminal "table", saved into a file,
   and subsequent splotting of the resulting grid data in that file works,
   but not completely as one expects it: (third plot in demo).
   In the data blocks of the generated file y is held constant, and x
   varies. Consequently, the plot is drawn "inside-out", as promised in
   sect. 38.1.2 of the manual.

Although this surely is no "bug" in a narrow sense, but I think it is not
as it should be, and may be sometimes annoying, at least irritating.

A phenomenon that may be related to this is probably a bug, otherwise an
error in the manual:

4) splotting binary data written as specified in sect. 38.1.1 results in
   both: interchanged x and y *meaning* and a plot with "inside-out" color
   choice. (Fourth plot in demo, provided your system uses IEEE
   representation of floats with big endians).
5) Both errors can be healed at once when plotting with "using 2:1:3"
   specifier: (fifth plot in demo). This may be a hint to the error.


Provided files (attached to this e-mail):
"xy_mixup.gpl":     gnuplot macro file for demonstration
"xy_griddata.gba":  ASCII griddata with correct specification
"xy_mixup.gba":     ASCII table data from gnuplot, generated by macro
"xy_mixup_bin.gbb": binary griddata file
                    (IEEE, big endian, generated on Intel PC)
"xy_mixup_bin.c":   C source for generating the binary griddata file

My system: MS-Windows 32 bit, or CYGWIN-NT 5.0 (X11)

Best regards,
               J. Prahl.

Majalee Aaditya wrote:

> but i feel there is one thing lacking, which the gnuplot 3.7, patchlevel
> 3 had. that is the "file completion" by making use of the "tab" key.

That feature is not missing in gnuplot-4.0, but only in your current 
compilation of it.  That's because it depends on which readline code you
compile in.  You need the GNU readline library to get filename completion.

hi,

after looking at the new things implemented in gnuplot 4.0, i was really
amazed. the main feature which amazed me was the mouse hot key concept.
i.e., zooming onto a particular area, and rotating the view of the
surface plot.

but i feel there is one thing lacking, which the gnuplot 3.7, patchlevel
3 had. that is the "file completion" by making use of the "tab" key.
i feel, if the same can be implemented, in gnuplot 4.0, it would just
make "gnuplot" invincible and it would be hands down for "matlab".

please do write back how to modify the source to get the "file
completion" going.

thanks.

regards,

aaditya


-- 

=====================================
 Majalee Aaditya Vasudev
 Scientist-SD
 Institute for Plasma Research   
 Near Indira Bridge, Bhat,
 Gandhinagar - 382 428
 Gujarat State
 INDIA
 Tel # : +91-79-2396 9031 (Ext:306)
 Fax # : +91-79-2396 9017
 email : ka...@ip...
=====================================