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

Re: square behavior

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

On Monday 10 January 2005 02:00 pm, Daniel J Sebald wrote:
> So, given what various people know about page layout, what does everyone 
> think about the way of approaching this?  A higher level cropping of the 
> image?  Or a more low-level crop for each terminal?

There's no way you could do it in the core code.

For some terminals, including PostScript, it is not possible to do it
at all, since the physical "pixels" occupied by characters are determined
by the font present in your ultimate output device.  Nobody upstream of
that can predict them perfectly.


-- 
Ethan A Merritt       merritt@u.washington.edu
Biomolecular Structure Center
Mailstop 357742
University of Washington, Seattle, WA 98195

Re: Fill plot area with background color

[Harald H. <h.h...@tu...>]

On Mon, 10 Jan 2005, Ethan Merritt wrote:

> > > > >In my company, most plots are
> > > > >filled by a yellow background
> >
> > > It is already possible for many terminal types.
> > > Which ones are you interested in?
> >
> > I am mainly using postscript.
>
> To recap my contribution to the previous discussion:
>
> (1) Think of an eps file as having a transparent background.
> gnuplot's PostScript output is agnostic with regard to
> the background color. Put it on any background you like when
> you import it into another document with specific layout or
> style requirements.

Yes, I know that possibility. But it is trial and error to get the right
position of the colored rectangle.

> (2) Adding a fully general capability is better than hacking
> specific work-arounds into the driver code.
> I have planned eventually to implement a new core command
>
>     set rectangle {n} <xlow>,<ylow>,<xhigh>,<yhigh>
>         {<fillstyle>} {front|back}
>
> The boundaries would accept either graph or plot coords,
> and you'd get the full range of fill-style colors, borders,
> and other options.

That idea is better then mine. I will see if I can find the time to
programme it. Should not be too complicated.

Harald

-- 
Harald Harders
h.h...@tu...
http://www.harald-harders.de

Re: "image"-related exotic interoperability question

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

Andreas K. Huettel wrote:

>I don't know if this is of interest at all to anyone else, but would it be
>possible without a lot of effort to force gnuplot to output a non-indexed
>rgb bitmap (with three times the size of course)? particularly since this 
>type of image is already produced by other plot commands...
>  
>

Does this seem like an option worth adding to the PostScript terminal?

That is, in PostScript we have image output that can be either indexed 
or a colorimage. The obvious connection is palette-based images are 
indexed, RGB images are colorimage. Would it be worthwhile to have an 
option that forces the type of output image to either always indexed or 
always colorimage? The immediate motivation here is a work-around for a 
problem that may or may not be the terminal drivers fault. I guess that 
may be reason enough, but is there a more legitimate reason for doing this?

It would be easy enough I think to generate a color image from an 
indexed image. Vice-versa might not work so well. The palette would be 
very large. So maybe an option of "indexed" for "set term postscript". 
I've no strong opinion... although I'd sort of like to stick with 
palette being indexed, RGB being color image.

Dan




-- 

Dan Sebald
email: daniel DOT sebald AT ieee DOT org
URL: http://acer-access DOT com/~dsebald AT acer-access DOT com/

new command for gnuplot.el: gnuplot-backslash-region

[James R. V. Z. <jr...@co...>]

--------------------------------------------------------------
Insert your description of the gnuplot-mode bug here.
Please be as specific as possible.

There are several known shortcomings of gnuplot-mode.
Many of these have to do with the complicated and inconsistent
syntax of gnuplot itself.  See the document string for the
function `gnuplot-mode' (use `C-h f') for details.

Note that this bug report form should be used for problems
with gnuplot-mode only.  Problems with gnuplot itself should
be addressed directly to the developers of gnuplot.
The maintainer of gnuplot-mode will not field questions about
gnuplot itself.  Thank you.
--------------------------------------------------------------

Bruce Ravel -

Thanks for writing gnuplot.el!  I assume you're still maintaining it.

gnuplot allows commands to be continued on following lines, but
gnuplot-mode doesn't automate backslashification (as it does in
cc-mode or make-mode).  I suggest the changes below, which I stole
from make-mode.

By the way, sh-mode should also automate backslashification, and there
are probably other languages that follow the same convention.  It
seems to me we need a generic backslash-region command that could be
invoked by any of the appropriate modes.  On the other hand, there are
some differences between c-backslash-region and
makefile-backslash-region which I could not understand and which
prevent a common implementation.  (I'm not really much of an elisp
programmer.  You'll probably want to check that the difference isn't
important for gnuplot scripts, or else that I picked the right version
by chance :-)

	- Jim Van Zandt

--- gnuplot.el	2004-04-10 10:42:33.000000000 -0400
+++ ../../../gnuplot.el	2005-01-10 17:15:13.000000000 -0500
@@ -633,6 +633,7 @@
   (define-key gnuplot-mode-map "\C-c\C-u" 'gnuplot-bug-report)
   (define-key gnuplot-mode-map "\C-c\C-v" 'gnuplot-send-line-and-forward)
   (define-key gnuplot-mode-map "\C-c\C-z" 'gnuplot-customize)
+  (define-key gnuplot-mode-map "\C-c\C-\\" 'gnuplot-backslash-region)
   (define-key gnuplot-mode-map "\M-\r"    'gnuplot-complete-keyword)
   (define-key gnuplot-mode-map "\M-\t"    'gnuplot-complete-keyword)
   (define-key gnuplot-mode-map "\C-i"     'indent-for-tab-command)
@@ -1137,6 +1138,16 @@
 		 (const :tag "No toolbar" nil))
   :group 'gnuplot-toolbar)
 
+(defcustom gnuplot-backslash-column 48
+  "*Column in which `gnuplot-backslash-region' inserts backslashes."
+  :type 'integer
+  :group 'gnuplot)
+
+(defcustom gnuplot-backslash-align t
+  "*If non-nil, `gnuplot-backslash-region' will align backslashes."
+  :type 'boolean
+  :group 'gnuplot)
+
 (defvar gnuplot-toolbar-location "")
 
 (defun gnuplot-toolbar-setup-toolbar (toolbar)
@@ -2150,6 +2161,76 @@
       (customize-group "gnuplot")
     (message "The Custom library is not installed.")))
 
+;; Backslashification.  Stolen from make-mode.el.
+
+(defun gnuplot-backslash-region (from to delete-flag)
+  "Insert, align, or delete end-of-line backslashes on the lines in the region.
+With no argument, inserts backslashes and aligns existing backslashes.
+With an argument, deletes the backslashes.
+
+This function does not modify the last line of the region if the region ends
+right at the start of the following line; it does not modify blank lines
+at the start of the region.  So you can put the region around an entire 
+gnuplot command and conveniently use this command."
+  (interactive "r\nP")
+  (save-excursion
+    (goto-char from)
+    (let ((column gnuplot-backslash-column)
+          (endmark (make-marker)))
+      (move-marker endmark to)
+      ;; Compute the smallest column number past the ends of all the lines.
+      (if gnuplot-backslash-align
+	  (progn
+	    (if (not delete-flag)
+		(while (< (point) to)
+		  (end-of-line)
+		  (if (= (preceding-char) ?\\)
+		      (progn (forward-char -1)
+			     (skip-chars-backward " \t")))
+		  (setq column (max column (1+ (current-column))))
+		  (forward-line 1)))
+	    ;; Adjust upward to a tab column, if that doesn't push
+	    ;; past the margin.
+	    (if (> (% column tab-width) 0)
+		(let ((adjusted (* (/ (+ column tab-width -1) tab-width)
+				   tab-width)))
+		  (if (< adjusted (window-width))
+		      (setq column adjusted))))))
+      ;; Don't modify blank lines at start of region.
+      (goto-char from)
+      (while (and (< (point) endmark) (eolp))
+        (forward-line 1))
+      ;; Add or remove backslashes on all the lines.
+      (while (and (< (point) endmark)
+                  ;; Don't backslashify the last line
+                  ;; if the region ends right at the start of the next line.
+                  (save-excursion
+                    (forward-line 1)
+                    (< (point) endmark)))
+        (if (not delete-flag)
+            (gnuplot-append-backslash column)
+          (gnuplot-delete-backslash))
+        (forward-line 1))
+      (move-marker endmark nil))))
+
+(defun gnuplot-append-backslash (column)
+  (end-of-line)
+  ;; Note that "\\\\" is needed to get one backslash.
+  (if (= (preceding-char) ?\\)
+      (progn (forward-char -1)
+             (delete-horizontal-space)
+             (indent-to column (if gnuplot-backslash-align nil 1)))
+    (indent-to column (if gnuplot-backslash-align nil 1))
+    (insert "\\")))
+
+(defun gnuplot-delete-backslash ()
+  (end-of-line)
+  (or (bolp)
+      (progn
+ 	(forward-char -1)
+ 	(if (looking-at "\\\\")
+ 	    (delete-region (1+ (point))
+ 			   (progn (skip-chars-backward " \t") (point)))))))
 
 
 ;;; --- help from the info file, keyword list + completion, insert function






Emacs  : GNU Emacs 21.3.1 (i386-pc-linux-gnu, X toolkit, Xaw3d scroll bars)
 of 2004-10-16 on raven, modified by Debian
Package: gnuplot-mode (version 0.6.0.vz)

current state:
==============
(setq
 gnuplot-mode-hook nil
 gnuplot-load-hook nil
 gnuplot-after-plot-hook '(gnuplot-trim-gnuplot-buffer)
 gnuplot-info-hook nil
 gnuplot-comint-setup-hook nil
 gnuplot-program "gnuplot"
 gnuplot-program-version "4.0"
 gnuplot-process-name "gnuplot"
 gnuplot-gnuplot-buffer "plot.gp"
 gnuplot-display-process 'window
 gnuplot-info-display 'window
 gnuplot-echo-command-line-flag t
 gnuplot-insertions-show-help-flag nil
 gnuplot-delay 0.01
 gnuplot-quote-character "'"
 gnuplot-keywords-when 'deferred
 gnuplot-toolbar-display-flag nil
 gnuplot-toolbar-use-toolbar nil
 gnuplot-gui-popup-flag nil
 gnuplot-gui-frame-plist '(height 18 width 65 border-width 0 user-position t
			   top 150 left 150 internal-border-width 0
			   unsplittable t default-toolbar-visible-p nil
			   has-modeline-p nil menubar-visible-p nil)
 gnuplot-gui-frame-parameters '((height . 18) (width . 65) (user-position . t)
				(top . 150) (left . 150) (border-width . 0)
				(menu-bar-lines . 0) (unsplittable . t))
 gnuplot-gui-fontname-list '(" " "\"Helvetica\"" "\"Times-Roman\"")
 gnuplot-gui-plot-splot-fit-style 'simple
 gnuplot-comint-recent-buffer #<buffer vel.dem>
 gnuplot-version "0.6.0.vz"
 Info-directory-list nil
 exec-path '("/usr/local/bin" "/usr/bin" "/bin" "/usr/bin/X11" "/usr/games"
	     "/home/jrv/bin" "/usr/local/netpbm" "/usr/bin/mh"
	     "/usr/lib/emacs/21.3/i386-linux")
 features '(sendmail reporter newcomment view gnuplot-gui cl cl-19 wid-edit
	    image gnuplot info-look info rmail compile mailcrypt rfc822 comint
	    tempo w3-cus url url-proxy url-privacy url-expand url-methods
	    url-history url-cookie url-util url-parse url-auto timezone
	    url-vars mailcap mail-parse rfc2045 rfc2047 gnus-util nnheader
	    mail-utils qp rfc2231 ietf-drums mm-util mule-util time-date
	    parse-time mail-prsvr xml-parse doxymacs debian-el
	    debian-el-loaddefs psgml-init mailcrypt-init dpkg-dev-el
	    dpkg-dev-el-loaddefs xcscope ring easymenu tex-site calc-init
	    disp-table ediff-hook vc-hooks lisp-float-type lisp-mode page
	    menu-bar timer select scroll-bar mldrag mouse facemenu faces
	    cus-face frame ucs-tables misc-lang vietnamese tibetan thai lao
	    korean japanese hebrew greek romanian slovak czech european
	    ethiopic devanagari indian cyrillic chinese case-table help
	    overlay md5 base64 format mule env backquote custom widget
	    x-toolkit)
 )

Re: Username patch

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

Juergen Wieferink wrote:

>On Sunday 09 January 2005 20:06, Daniel J Sebald wrote:
>  
>
>>For some reason "ggv" and "xpdf" do not show file information.  Why?  I
>>don't know.
>>    
>>
>
>Xpdf is distributed with the command line tool pdfinfo.
>  
>

Well, I'll be.  The displayed info through pdfinfo is the same as shown 
in Acroread.  So, the patch seems to work correctly as far as I can tell.

Dan

Re: square behavior

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

Ethan Merritt wrote:

>On Monday 10 January 2005 11:48 am, Daniel J Sebald wrote:
>  
>
>>>For the png driver there is the additional option of using
>>>
>>>	set term png crop
>>>      
>>>
>>YES!!!!!!!!  Everything fine trimmed right up to the first non-white 
>>pixels.  Perfect.  Every terminal should have such an option.
>>    
>>
>
>Feel free to contribute one for your favorite driver.
>

Well, this is what I'm wondering.  Is this feature as straightforward 
in, say, PostScript as it might be in PNG?  I haven't looked, but I 
imagine PNG , being pixel-based, can easily tell where it's limits are.  
But with PostScript, you have plot layout in terms of elements like 
fonts, lines, etc.  That might be a more difficult problem, and in fact 
it might be easier to do the cropping at a higher level of gnuplot's 
plot layout.  Heck, couldn't the word "crop" be worked into one of the 
plotting commands.

So, given what various people know about page layout, what does everyone 
think about the way of approaching this?  A higher level cropping of the 
image?  Or a more low-level crop for each terminal?

Dan

Command line way to prevent unbuffered?

[Jim K. <je...@kl...>]

I'm using gnuplot to create PDF files most of the time.  I find that the 
unbuffered setting to make X11 work with the mouse and pasteboard vastly 
increases the syscall overhead for this batch operation.  Would it make 
sense to add an argument to be able to permit buffering?  I'm using 
cygwin so perhaps there is some set of porting options that interact on 
this platform.

Thanks - Jim

Re: Fill plot area with background color

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

On Monday 10 January 2005 11:05 am, Harald Harders wrote:
> Is there a way to fill the plot area with a background color?
> What I mean is similar to this example:
> 
> set terminal postscript eps color rounded linewidth 2
> set output 'asdf.eps'
> set style line 1 lt 1 lw 4
> set style line 9999 lt 6 lw 1
> set style fill solid 1
> plot 1 with filledcurves above y1=-1 ls 9999,\
>      sin(x) ls 1
> 
> Unfortunately, the tics are lost here.

To get your tics back, add the command
  set grid front nox noy noz


-- 
Ethan A Merritt       merritt@u.washington.edu
Biomolecular Structure Center
Mailstop 357742
University of Washington, Seattle, WA 98195

Re: Fill plot area with background color

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

On Monday 10 January 2005 11:45 am, Harald Harders wrote:
> > > Harald Harders wrote:
> > >
> > > >In my company, most plots are
> > > >filled by a yellow background 
> 
> > It is already possible for many terminal types.
> > Which ones are you interested in?
> 
> I am mainly using postscript.

To recap my contribution to the previous discussion: 

(1) Think of an eps file as having a transparent background.
gnuplot's PostScript output is agnostic with regard to
the background color. Put it on any background you like when
you import it into another document with specific layout or
style requirements.
 
(2) Adding a fully general capability is better than hacking
specific work-arounds into the driver code.
I have planned eventually to implement a new core command

    set rectangle {n} <xlow>,<ylow>,<xhigh>,<yhigh>
        {<fillstyle>} {front|back}

The boundaries would accept either graph or plot coords,
and you'd get the full range of fill-style colors, borders,
and other options.

I won't have time to work on this anytime soon, however,
so if you'd like to plunge in and contribute a patch 
I'll help refine and polish it when I emerge from the end
of my current press of deadlines.

-- 
Ethan A Merritt       merritt@u.washington.edu
Biomolecular Structure Center
Mailstop 357742
University of Washington, Seattle, WA 98195

Re: Username patch

[Juergen W. <wie...@fr...>]

On Sunday 09 January 2005 20:06, Daniel J Sebald wrote:
> For some reason "ggv" and "xpdf" do not show file information.  Why?  I
> don't know.

Xpdf is distributed with the command line tool pdfinfo.


Juergen

Re: square behavior

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

On Monday 10 January 2005 11:48 am, Daniel J Sebald wrote:
> 
> >For the png driver there is the additional option of using
> >
> >	set term png crop
> 
> YES!!!!!!!!  Everything fine trimmed right up to the first non-white 
> pixels.  Perfect.  Every terminal should have such an option.

Feel free to contribute one for your favorite driver.


-- 
Ethan A Merritt       merritt@u.washington.edu
Biomolecular Structure Center
Mailstop 357742
University of Washington, Seattle, WA 98195

Re: square behavior

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

Ethan Merritt wrote:

>On Monday 10 January 2005 10:38 am, Daniel J Sebald wrote:
>  
>
>>No, I meant the "set size square" problem.  Gnuplot creates a plot that 
>>is square.  However, the whitespace it leaves is still the same aspect 
>>ratio as the default.
>>    
>>
>
>I would claim that is the correct thing for it to do.
>
>The size of the "page" should be set at the terminal level.
>The size and aspect ratio of the "plot" should be set by
>the core command "set size".   This is how it works for
>multiplot, right?
>  
>

I suppose you are correct.  However, I think we are left with the same 
trial-and-error-ish-ness.

>If you change things so that "set size" affects the whole
>page size, then won't it totally break multiplot?
>
>If you want a square plot on a square page, then you need to do
>something like:
>
>	set term png size 500,500
>	set size square
>  
>

PostScript doesn't have a size (or more appropriately BoundingBox) 
specification.  But still, what would be the difference of having to 
tweak the values in the PostScript file or at the "set term" line?  In 
the former case at least you can just modify the number and reload the 
PS file in Ghostview.  In the latter case one must continually reissue 
the plotting sequence of commands, then reload the PS file in Ghostview.

I just tried an example with the above PNG syntax.  With a color box on 
the right side, and the square image, and the 500 by 500 output, there 
remains whitespace now at the top of the image.

>For the png driver there is the additional option of using
>
>	set term png crop
>	set size square
>
>which may be closer to what you are looking for.
>  
>

YES!!!!!!!!  Everything fine trimmed right up to the first non-white 
pixels.  Perfect.  Every terminal should have such an option.

Dan

Re: Fill plot area with background color

[Harald H. <h.h...@tu...>]

On Mon, 10 Jan 2005, Ethan Merritt wrote:

> On Monday 10 January 2005 11:26 am, Daniel J Sebald wrote:
> > Harald Harders wrote:
> >
> > >Unfortunately, the tics are lost here. In my company, most plots are
> > >filled by a yellow background (it does not matter if I like it). Now the
> > >question is if I should programme a patch that provides something like
> > >
> > >set backgroundcolor graph #rgb screen #rgb
> > >
> >
> > On 11/28/04 thru 12/05/04 was a discussion about this. Petr proposed a
> > similar, but slightly different syntax. Ethan had some comments prior to
> > that about the functionality possibly already being present.

Strange that I do not remember.

> It is already possible for many terminal types.
> Which ones are you interested in?

I am mainly using postscript.

-- 
Harald Harders
h.h...@tu...
http://www.harald-harders.de

Re: Fill plot area with background color

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

On Monday 10 January 2005 11:26 am, Daniel J Sebald wrote:
> Harald Harders wrote:
> 
> >Unfortunately, the tics are lost here. In my company, most plots are
> >filled by a yellow background (it does not matter if I like it). Now the
> >question is if I should programme a patch that provides something like
> >
> >set backgroundcolor graph #rgb screen #rgb
> >
> 
> On 11/28/04 thru 12/05/04 was a discussion about this. Petr proposed a 
> similar, but slightly different syntax. Ethan had some comments prior to 
> that about the functionality possibly already being present.

It is already possible for many terminal types.
Which ones are you interested in?



-- 
Ethan A Merritt       merritt@u.washington.edu
Biomolecular Structure Center
Mailstop 357742
University of Washington, Seattle, WA 98195

Re: Fill plot area with background color

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

Harald Harders wrote:

>Unfortunately, the tics are lost here. In my company, most plots are
>filled by a yellow background (it does not matter if I like it). Now the
>question is if I should programme a patch that provides something like
>
>set backgroundcolor graph #rgb screen #rgb
>  
>

On 11/28/04 thru 12/05/04 was a discussion about this. Petr proposed a 
similar, but slightly different syntax. Ethan had some comments prior to 
that about the functionality possibly already being present.

Dan

Re: square behavior

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

On Monday 10 January 2005 10:38 am, Daniel J Sebald wrote:
> 
> No, I meant the "set size square" problem.  Gnuplot creates a plot that 
> is square.  However, the whitespace it leaves is still the same aspect 
> ratio as the default.

I would claim that is the correct thing for it to do.

The size of the "page" should be set at the terminal level.
The size and aspect ratio of the "plot" should be set by
the core command "set size".   This is how it works for
multiplot, right?

If you change things so that "set size" affects the whole
page size, then won't it totally break multiplot?

If you want a square plot on a square page, then you need to do
something like:

	set term png size 500,500
	set size square

For the png driver there is the additional option of using

	set term png crop
	set size square

which may be closer to what you are looking for.


-- 
Ethan A Merritt       merritt@u.washington.edu
Biomolecular Structure Center
Mailstop 357742
University of Washington, Seattle, WA 98195

Fill plot area with background color

[Harald H. <h.h...@tu...>]

Is there a way to fill the plot area with a background color?
What I mean is similar to this example:

set terminal postscript eps color rounded linewidth 2
set output 'asdf.eps'
set style line 1 lt 1 lw 4
set style line 9999 lt 6 lw 1
set style fill solid 1
plot 1 with filledcurves above y1=-1 ls 9999,\
     sin(x) ls 1

Unfortunately, the tics are lost here. In my company, most plots are
filled by a yellow background (it does not matter if I like it). Now the
question is if I should programme a patch that provides something like

set backgroundcolor graph #rgb screen #rgb

or if there already is a possibility to reach that.

Greetings
Harald

-- 
Harald Harders
h.h...@tu...
http://www.harald-harders.de

Re: square behavior

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

Petr Mikulik wrote:

>>I've been using gnuplot quite a bit lately.  It's been working very
>>well, which says a lot about the amount of work put in over the last few
>>years.
>>
>>The only annoying thing right now is the layout for plots when in square
>>mode.  It's easy enough to modify the bounding box in an EPS file, but
>>it sure would be nice if some brave soul could slay that dragon.
>>    
>>
>
>You mean "with image"? I think that the current bounding box calculator does
>not include the colour box.
>

No, I meant the "set size square" problem.  Gnuplot creates a plot that 
is square.  However, the whitespace it leaves is still the same aspect 
ratio as the default.  Hence there is a chunk of whitespace on the right 
side.

A while back I put a patch on SF that balances the white space on both 
sides.  It looks better, but actually--as Ethan pointed out--it's hardly 
an improvement and there is a proper way of correcting it.   (One has to 
modify two borders in the PostScript file, or whatnot, rather than one.)

Dan

Re: square behavior

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

> I've been using gnuplot quite a bit lately.  It's been working very
> well, which says a lot about the amount of work put in over the last few
> years.
>
> The only annoying thing right now is the layout for plots when in square
> mode.  It's easy enough to modify the bounding box in an EPS file, but
> it sure would be nice if some brave soul could slay that dragon.

You mean "with image"? I think that the current bounding box calculator does
not include the colour box.

--
PM

Re: "image"-related exotic interoperability question

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

> I'm going to send you a patch file for correcting a problem with scaling
> that Andreas found in the image routines for splot mode.  (The splot was
> using the outer limits for plot, hence the images were coming out bigger
> than expected.)  There is also a better routine for determining what
> portion of the image is within the range volume of the splot.  Please
> apply the patch as one or two bug fixes.

It has been applied.

---
PM

Re: Open patches that are useful for me

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

> #616199 Ignore title string after notitle
> #713166 margins changed from int to float
> #835235 minitics updates

All these patches are useful, I support commiting them.

---
PM

CorelDraw blank images

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

Daniel J Sebald wrote:

> I've done a web search and can't find any cases of CorelDraw users 
> reporting such behavior.


I'm just curious if it is a problem with gnuplot images.

CorelDraw blank images

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

Anyone familiar with CorelDraw?  Someone's having a problem importing a 
gnuplot PostScript image of indexed palette type.  (Apparently the color 
images import fine.)  The result in CorelDraw is a black rectangle.  The 
PostScript image remains visible in stuff like Ghostview, GIMP, etc.  
The gnuplot syntax for indexed images is below.  I've done a web search 
and can't find any cases of CorelDraw users reporting such behavior.

Dan


%%BeginImage
gsave
987 3853 translate
5562 -3326 scale
%%BeginPalette
[ /Indexed
  /DeviceRGB 255
  <
   000000 100006 17000d 1c0013 200019 24001f 270026 2a002c
<snip>
   fbeb00 fced00 fcf000 fdf300 fdf600 fef900 fefc00 ffff00
  >
] setcolorspace
%%EndPalette
<<
  /ImageType 1
  /Width 128
  /Height 128
  /BitsPerComponent 8
  /ImageMatrix [ 128 0 0 128 0 0 ]
  /Decode [ 0 255 ]
  /imagebuf 16384 string def
  /DataSource {currentfile imagebuf readhexstring pop}
  /MultipleDataSources false
  /Interpolate false
 >>
image
e8e8e8e8e8e8e8e9e9e9e9e9e9e9e9e9e9e9e9e9e9e9e9e9e9e9e9e9e9e9e9e9e9e9e9e9e9e9e9
<snip>
81818181
grestore
%%EndImage

square behavior

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

I've been using gnuplot quite a bit lately.  It's been working very 
well, which says a lot about the amount of work put in over the last few 
years.

The only annoying thing right now is the layout for plots when in square 
mode.  It's easy enough to modify the bounding box in an EPS file, but 
it sure would be nice if some brave soul could slay that dragon.

Dan

Re: "image"-related exotic interoperability question

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

Petr,

I'm going to send you a patch file for correcting a problem with scaling 
that Andreas found in the image routines for splot mode.  (The splot was 
using the outer limits for plot, hence the images were coming out bigger 
than expected.)  There is also a better routine for determining what 
portion of the image is within the range volume of the splot.  Please 
apply the patch as one or two bug fixes.

Also, in looking at a problem with PostScript palettes (one still 
unresolved but may be a problem with the application and not Gnuplot), I 
was reminded of a patch that Ethan and I created for better handling 
Level1 and Level2 PostScripting.  It doesn't seem like that was moved in 
yet.  We didn't have access to a PostScript Level 1 only application so 
technically haven't tested it.  However, I think what was in the patch 
was an improvement.  It was a fairly big patch (not more code, big 
because of reshuffling of code), so the risk of an unforseen bug is 
probably less than forgetting about the patch some day.  Ethan, should 
we move that into CVS?  (The latest date I have is 
"image_post_19sep2004.patch.gz".)

Dan