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

Re: Lua/Tikz terminal behavior

[sfeam <sf...@us...>]

On Saturday, 08 November 2014 07:12:43 AM Philipp K. Janert wrote:
> 
> Thanks all for your comments.
> I am still looking for the advantages over,
> say, the epslatex terminal.

Specifically in comparison to epslatex
1) epslatex produces *.eps, not *.pdf, so not as nice if you use pdflatex
2) eoslatex does not support transparency or alpha channel images.

A more equal comparison would be to 'set term cairolatex pdf'

	Ethan

Re: RC2 : Problems with multiplot ... spacing

[<pl...@pi...>]

On 11/07/14 16:35, Philipp K. Janert wrote:
> On Fri, 07 Nov 2014 10:55:43 +0100
> Christoph Bersch <us...@be...> wrote:
>
>> Zitat von "Philipp K. Janert" <ja...@ie...>:
>>>
>>> 1) The new (?) "spacing" option is not documented
>>> at all. I can vaguely guess what it does, but not
>>> entirely.
>>
>> There is a small paragraph including an example at the end of the
>> `set multiplot` section in the documentation.
>>
>> The `spacing` option was thought to be used together with the
>> `margins` option. The `margins` option sets all outer margins of the
>> multiplot area, whereas `spacing` gives the vertical and/or
>> horizontal spacing between two plots.
>
> Yes, I saw that paragraph, but it did leave me a
> little in the dark. I guess "spacing" is using the
> same units as "margins"?
>
> [snip]
>
>>
>> Also: I'm not sure, if had had thought of using `spacing` without
>> also setting `margins` and if this is possible. In any case I'll
>> check if using `spacing` alone could work or insert an appropriate
>> error message it one of the two options isn't set.
>>
>
> I don't think it has anything to do with the
> concurrent use (or non-use) or "margin".
>
> It is also weird that gnuplot gets itself so
> wedged that even a "reset" does not help - I
> have to quit and restart after every
> "mulitplot ... spacing".
>
> (I really like the feature, otherwise - it solves
> a recurrent problem with multiplot.)
>
> Best,
>
> 		Ph.
>

Yes, this is worth getting working correctly and fully documenting. I 
long ago gave up using multiplot due to the amount of real estate that 
it wasted in handling margins and spacing around graphs.

I've usually ended up doing collation in a graphics processing tool on 
individual gnuplot graphs.

There was a lot of grumbling from Octave devs about this too, who use gp 
as a plotting backend.

Peter.

Re: Lua/Tikz terminal behavior

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

Thanks all for your comments.

Mojca - you are absolutely right: I forgot
to close the files. 
	replot; set o
did the trick.

Also thanks for the advice on quoting/escaping,
that works now, too.

I saw the bug report on arrows, so that's 
known behavior.

I am still looking for the advantages over,
say, the epslatex terminal.

Best,

		Ph.

Re: Fw: Lua/Tikz terminal behavior

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

On Sat, Nov 8, 2014 at 7:25 AM, Philipp K. Janert wrote:
>
> Is there a way to verify that, at runtime, gnuplot
> is using the lua files from that install dir? And
> that Latex (when I later compile the resulting Tex
> file) is finding the correct version of these files?

I don't know the answer to the first question, but the second answer
is yes. Check the output written by TeX:

This is pdfTeX, Version 3.14159265-2.6-1.40.15 (TeX Live 2014)
(preloaded format=pdflatex)
 restricted \write18 enabled.
entering extended mode
(./t7.tex
LaTeX2e <2014/05/01>
Babel <3.9k> and hyphenation patterns for 2 languages loaded.
(/opt/local/share/texmf-texlive/tex/latex/base/article.cls
Document Class: article 2007/10/19 v1.4h Standard LaTeX document class
(/opt/local/share/texmf-texlive/tex/latex/base/size10.clo))
[...]
(/opt/local/share/texmf/tex/latex/gnuplot-lua-tikz/gnuplot-lua-tikz.sty
(/opt/local/share/texmf/tex/latex/pgf/frontendlayer/tikz.sty
[...]

Mojca

Re: Lua/Tikz terminal behavior

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

On Sat, Nov 8, 2014 at 4:13 AM, Philipp K. Janert wrote:
>
> How much experience do people have with the Lua/Tikz
> terminal? I have been experimenting with it a bit,
> and I found some of the behavior bizarre (not to
> mention obscure).

This won't solve all of the problems, but you need to make sure that
you finish writing to a file before running TeX.

> Below is a log of various sessions. Three problems
> that I seem to have narrowed down: when I try to
> write successively to more than one file during one
> gnuplot session, the subsequent tex files do not
> compile. Tex stops at some point, waiting for input.
> (See session 2 below.)

Try compiling 't3.tex' again. I'm almost sure that it will work now.
Please check the contents of the file before compiling it. Either exit
gnuplot, change the terminal or run 'set o[utput]' before compiling
the tex file.

> When using arrows, gnuplot fails to create the proper
> lua script (which is bad), but when I then try to
> exit gnuplot, it gets completely wedged, continuously
> spewing messages about "Missing Lua context". One has
> to kill it from a separate shell. (Session 3)
>
> I can use arrows when I set the terminal option
> "gparrows". I tried this after seeing something
> about "misusing the angle value" in the doc. I am
> not sure, though, what the doc is really saying -
> what should I do if I want to use Lua/Tikz-arrows?

There must be some arrow-related bug somewhere. I suspect that the
number of people using both arrows and tikz could be small enough that
nobody noticed the problem so far.

> Finally, how do I include Tex constructs in text labels

Just like you would expect (except that you need to use proper quotes
or escapes, see below).

> (which, I assume, is the whole point). The documentation
> does not seem to be mention that at all. I tried the
> same syntax as in the epslatex terminal, namely simply
> including Tex code:
>         set label "Hello $\int e^{-x}$" at 0,0
> but this only led to one of those tex files that
> don't compile. (Session 6)

Just because you didn't finish the file.

> At this point, I find it all rather obscure. I also
> wonder whether it is worthwhile trying to figure this
> out - what can this terminal do that none of the
> others can do? I searched the web a little and did
> not find much.
>
> Any experiences, insights, opinions?
>
> Best,
>
>                 Ph.
>
>
> SESSION TRANSCRIPTS FOLLOW
>
>
> SESSION 1
>
> plot sin(x)
> set t tikz standalone
> set o 't1.tex'
> replot
> set o
>
> shell> pdflatex t1.tex : OK
>
> ============================================================
> SESSION 2
>
> plot sin(x)
> set t tikz standalone
> set o 't2.tex'
> replot
> set o
> set o 't3.tex'
> replot
>
> shell> pdflatex t2.tex : OK
> shell> pdflatex t3.tex :
>
> Preview: Fontsize 10pt
> Preview: PDFoutput 1
>
> Overfull \hbox (10.65929pt too wide) in paragraph at lines 1083--1083
> [][]
> Preview: Tightpage 0 0 0 0
> [1{/var/lib/texmf/fonts/map/pdftex/updmap/pdftex.map}])
> *

You tried to compile unfinished file.

> ============================================================
> SESSION 3
>
> plot sin(x)
> set arrow from -5,-.4 to 2,.6
> set t tikz standalone
> set o 't4.tex'
> replot
>          .../rc52/staging/share/gnuplot/5.0/lua/gnuplot-tikz.lua:346:
> attempt to concatenate local 'direction' (a nil value) stac

I get the same problem. This looks like a bug in gnuplot-tikz.lua.

> BUT THEN:
>
> I wish to quit gnuplot using ^D, but it doesn't terminate,
> instead it prints FOREVER (unstoppable, must close window):

Side effect of the lua script taking over everything and failing to do
the right thing.

> ============================================================
> SESSION 6
>
> plot sin(x)
> set label "Hello $\int e^{-x}$" at 0,0

You are making one mistake here. You should use either single quotes:
    'Hello $\int e^{-x}$'
or you should escape the backslash.

Please check the output in the resulting tex file before you compile
to make sure that the label is correct.

> set t tikz standalone
> set o 't7.tex'
> replot
>
> shell> pdflatex t7.tex :
>
> Preview: Fontsize 10pt
> Preview: PDFoutput 1
>
> Overfull \hbox (10.65929pt too wide) in paragraph at lines 561--561
> [][]
> Preview: Tightpage 0 0 0 0
> [1{/var/lib/texmf/fonts/map/pdftex/updmap/pdftex.map}])
> *

I strongly suspect that you are experiencing the same problem as
elsewhere: you tried to compile unfinished file. The example works for
me (I tested with both gnuplot 4.6.6 as well as with 5.0.rc2).

Mojca

Re: Lua/Tikz terminal behavior

[Christoph B. <us...@be...>]

Am 08.11.2014 04:13, schrieb Philipp K. Janert:
>
> I can use arrows when I set the terminal option
> "gparrows". I tried this after seeing something
> about "misusing the angle value" in the doc. I am
> not sure, though, what the doc is really saying -
> what should I do if I want to use Lua/Tikz-arrows?

There had been problems with some changes in the general arrow drawing 
code. That is fixed in the 5.0 branch but wasn't yet included in the 
rc2: http://sourceforge.net/p/gnuplot/bugs/1476/ (at the moment I cannot 
find the related post in some other bug or on the mailing list 
concerning the lua terminal). There have also been other additions and 
fixes to the lua/tikz terminal after RC2.

> Finally, how do I include Tex constructs in text labels
> (which, I assume, is the whole point). The documentation
> does not seem to be mention that at all. I tried the
> same syntax as in the epslatex terminal, namely simply
> including Tex code:
> 	set label "Hello $\int e^{-x}$" at 0,0
> but this only led to one of those tex files that
> don't compile. (Session 6)

For me that file compiles, but gives the wrong results since you don't 
escape the backslash. The following works fine for me (current 5.0 branch)

plot sin(x)
set label "Hello $\\int e^{-x}$" at 0,0
set t tikz standalone
set o 't7.tex'
replot

> plot sin(x)
> set arrow from -5,-.4 to 2,.6
> set t tikz standalone
> set o 't4.tex'
> replot
>           .../rc52/staging/share/gnuplot/5.0/lua/gnuplot-tikz.lua:346:
> attempt to concatenate local 'direction' (a nil value) stac
>
> BUT THEN:
>
> I wish to quit gnuplot using ^D, but it doesn't terminate,
> instead it prints FOREVER (unstoppable, must close window):
>
>      Missing Lua context! No script?

That error with the arrows is fixed. However, the general problem 
remains, that gnuplot crashes when an error in the lua script appears. I 
have no idea how to fix this.

I haven't worked much with the tikz terminal, but made some additions 
and fixes recently.

The mention of 'misuse of the angle parameter' is about using the angle 
parameter as index for an own arrow style which isn't in the list of 
predefined ones.

I do also have to find out some 'outstanding' features of the tikz 
terminal. As far as I know you can also define custom line styles, which 
now shouldn't be so outstanding, but before adding custom dash styles it 
was.

Christoph

Fw: Lua/Tikz terminal behavior

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

Ethan pointed out that I may be suffering from 
version skew between my executable (which is
straight RC2), and the various auxiliary files
that Lua/Tikz need:

> I suspect that your gnuplot executable is newer than the
> lua/gnuplot support scripts for TeX.  Make sure that these files
> have been updated during installation of your new gnuplot:
> 
> /usr/share/texmf-local/tex/latex/gnuplot/gnuplot-lua-tikz-common.tex
> /usr/share/texmf-local/tex/latex/gnuplot/gnuplot-lua-tikz.sty
> /usr/share/texmf-local/tex/latex/gnuplot/gnuplot-lua-tikz.tex
> /usr/share/texmf-local/tex/latex/gnuplot/t-gnuplot-lua-tikz.tex
> /usr/local/share/gnuplot/5.0/lua/gnuplot-tikz.lua
> 

I am working strictly with RC2, which has 
been installed (using "make install") into
some local directory (./configure --prefix...)  
The auxiliary files are all under that install
directory (and then /share/texmf/tex/latex/gnuplot).

Is there a way to verify that, at runtime, gnuplot
is using the lua files from that install dir? And
that Latex (when I later compile the resulting Tex
file) is finding the correct version of these files?

Best,

		Ph.

Lua/Tikz terminal behavior

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

How much experience do people have with the Lua/Tikz
terminal? I have been experimenting with it a bit,
and I found some of the behavior bizarre (not to 
mention obscure).

Below is a log of various sessions. Three problems
that I seem to have narrowed down: when I try to 
write successively to more than one file during one 
gnuplot session, the subsequent tex files do not
compile. Tex stops at some point, waiting for input.
(See session 2 below.)

When using arrows, gnuplot fails to create the proper
lua script (which is bad), but when I then try to
exit gnuplot, it gets completely wedged, continuously
spewing messages about "Missing Lua context". One has
to kill it from a separate shell. (Session 3)

I can use arrows when I set the terminal option
"gparrows". I tried this after seeing something
about "misusing the angle value" in the doc. I am
not sure, though, what the doc is really saying - 
what should I do if I want to use Lua/Tikz-arrows?

Finally, how do I include Tex constructs in text labels
(which, I assume, is the whole point). The documentation
does not seem to be mention that at all. I tried the
same syntax as in the epslatex terminal, namely simply
including Tex code:
	set label "Hello $\int e^{-x}$" at 0,0
but this only led to one of those tex files that
don't compile. (Session 6)

At this point, I find it all rather obscure. I also
wonder whether it is worthwhile trying to figure this
out - what can this terminal do that none of the 
others can do? I searched the web a little and did 
not find much.

Any experiences, insights, opinions?

Best,

		Ph.


SESSION TRANSCRIPTS FOLLOW


SESSION 1

plot sin(x)
set t tikz standalone
set o 't1.tex'
replot
set o

shell> pdflatex t1.tex : OK

============================================================
SESSION 2

plot sin(x)
set t tikz standalone
set o 't2.tex'
replot
set o
set o 't3.tex'
replot

shell> pdflatex t2.tex : OK
shell> pdflatex t3.tex : 

Preview: Fontsize 10pt
Preview: PDFoutput 1

Overfull \hbox (10.65929pt too wide) in paragraph at lines 1083--1083
[][] 
Preview: Tightpage 0 0 0 0
[1{/var/lib/texmf/fonts/map/pdftex/updmap/pdftex.map}])
*

============================================================
SESSION 3

plot sin(x)
set arrow from -5,-.4 to 2,.6
set t tikz standalone
set o 't4.tex'
replot
         .../rc52/staging/share/gnuplot/5.0/lua/gnuplot-tikz.lua:346:
attempt to concatenate local 'direction' (a nil value) stac

BUT THEN:

I wish to quit gnuplot using ^D, but it doesn't terminate,
instead it prints FOREVER (unstoppable, must close window):

    Missing Lua context! No script?

============================================================
SESSION 4

plot sin(x)
set arrow from -5,-.4 to 2,.6
set t tikz standalone gparrows
set o 't5.tex'
replot

shell> pdflatex t5.tex : OK

============================================================
SESSION 5

plot sin(x)
set label "Hello"
set t tikz standalone
set o 't6.tex'
replot

shell> pdflatex t6.tex : OK

============================================================
SESSION 6

plot sin(x)
set label "Hello $\int e^{-x}$" at 0,0
set t tikz standalone
set o 't7.tex'
replot

shell> pdflatex t7.tex :

Preview: Fontsize 10pt
Preview: PDFoutput 1

Overfull \hbox (10.65929pt too wide) in paragraph at lines 561--561
[][] 
Preview: Tightpage 0 0 0 0
[1{/var/lib/texmf/fonts/map/pdftex/updmap/pdftex.map}])
*

Re: RC2 : Problems with multiplot ... spacing

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

On Fri, 07 Nov 2014 10:55:43 +0100
Christoph Bersch <us...@be...> wrote:

> Zitat von "Philipp K. Janert" <ja...@ie...>:
> >
> > 1) The new (?) "spacing" option is not documented
> > at all. I can vaguely guess what it does, but not
> > entirely.
> 
> There is a small paragraph including an example at the end of the
> `set multiplot` section in the documentation.
> 
> The `spacing` option was thought to be used together with the  
> `margins` option. The `margins` option sets all outer margins of the  
> multiplot area, whereas `spacing` gives the vertical and/or
> horizontal spacing between two plots.

Yes, I saw that paragraph, but it did leave me a
little in the dark. I guess "spacing" is using the
same units as "margins"? 

[snip]

> 
> Also: I'm not sure, if had had thought of using `spacing` without
> also setting `margins` and if this is possible. In any case I'll
> check if using `spacing` alone could work or insert an appropriate
> error message it one of the two options isn't set.
> 

I don't think it has anything to do with the 
concurrent use (or non-use) or "margin".

It is also weird that gnuplot gets itself so
wedged that even a "reset" does not help - I
have to quit and restart after every 
"mulitplot ... spacing".

(I really like the feature, otherwise - it solves
a recurrent problem with multiplot.)

Best,

		Ph.

Re: Regression problem with gp5-rc2: make pdf fails

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

[snip]

> Using minipage and includegraphics should work fine, or also using
> the sidecap package and giving the images a caption on the side,
> when necessary. This would however required additional changes in
> the gnuplot.doc file to have more parts which contains specific
> formatting for the LaTeX output.
> 
> I can see if I can come up with a reasonable alternative.

It's not a big deal, since the pre-built pdf is 
part of the release. 

Maybe it's enough to make the dependency on a 
"non-standard" package clearer?

> 
> Christoph
> 
> 
> ------------------------------------------------------------------------------
> _______________________________________________
> gnuplot-beta mailing list
> gnu...@li...
> Membership management via:
> https://lists.sourceforge.net/lists/listinfo/gnuplot-beta

Re: Regression problem with gp5-rc2: make pdf fails

[Christoph B. <us...@be...>]

Zitat von Allin Cottrell <cot...@wf...>:
>
> As I understand Ethan's comment, he's talking about flowing text
> around an included graphic. I find I'm able to do this using
> the minipage environment plus \includegraphics from the
> graphicx package.

Indeed, the picins package is for flowing text around a graphic.  
Unfortunately, its license doesn't allow it to be included in a TeX  
distribution (at least not in TeXLive), which is the main problem.

I had tried replacing picins with wrapfigure, but didn't succeed. I  
couldn't get the settings correct to reproduce the current behavior.  
Especially paragraphs and page breaks mess things up.

Using minipage and includegraphics should work fine, or also using the  
sidecap package and giving the images a caption on the side, when  
necessary. This would however required additional changes in the  
gnuplot.doc file to have more parts which contains specific formatting  
for the LaTeX output.

I can see if I can come up with a reasonable alternative.

Christoph

Re: RC2 : Problems with multiplot ... spacing

[Christoph B. <us...@be...>]

Zitat von "Philipp K. Janert" <ja...@ie...>:
>
> 1) The new (?) "spacing" option is not documented
> at all. I can vaguely guess what it does, but not
> entirely.

There is a small paragraph including an example at the end of the `set  
multiplot` section in the documentation.

The `spacing` option was thought to be used together with the  
`margins` option. The `margins` option sets all outer margins of the  
multiplot area, whereas `spacing` gives the vertical and/or horizontal  
spacing between two plots.

Unfortunatly, the example as given in the documentation:

set multiplot layout 2,2 columnsfirst margins 0.1,0.9,0.1,0.9 spacing 0.1
set ylabel 'ylabel'
plot sin(x)
set xlabel 'xlabel'
plot cos(x)
unset ylabel
unset xlabel
plot sin(2*x)
set xlabel 'xlabel'
plot cos(2*x)
unset multiplot

doesn't work anymore (it used when the patch was checked in). That  
must have been broken.

Also: I'm not sure, if had had thought of using `spacing` without also  
setting `margins` and if this is possible. In any case I'll check if  
using `spacing` alone could work or insert an appropriate error  
message it one of the two options isn't set.


Thanks for testing!

Christoph

RC2 : Problems with multiplot ... spacing

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

Two issues:

1) The new (?) "spacing" option is not documented 
at all. I can vaguely guess what it does, but not
entirely. 

2) Using "spacing" repeatedly causes problems.

Consider this set of commands

set multiplot layout 2,1 spacing 0.1
plot sin(x)
plot cos(x)
unset multi
reset

If I put them into a file and then load them, the
graph comes up just fine. If I then try to load 
them again, the graph is blank, and I get the
following error msg:

         "m.gp", line 3: warning: Terminal canvas area too small to
hold plot. Check plot boundary and font sizes.
         "m.gp", line 4: warning: Terminal canvas area too small to
hold plot. Check plot boundary and font sizes.

BUT if I remove the "spacing 0.1" option from the
multiplot command, I can load the file repeatedly,
without any problems. 

Observed with RC2 and the wxt terminal.

Re: Regression problem with gp5-rc2: make pdf fails

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

On Thu, 6 Nov 2014 20:54:47 -0500 (EST)
Allin Cottrell <cot...@wf...> wrote:

> On Thu, 6 Nov 2014, Philipp K. Janert wrote:
> 
> > On Thu, 6 Nov 2014 20:16:30 -0500 (EST)
> > Allin Cottrell <cot...@wf...> wrote:
> >
> >> On Thu, 6 Nov 2014, Philipp K. Janert wrote:
> >>
> >>>
> >>> I seem to have a problem building the PDF of
> >>> the doc with RC2.
> >>>
> >>> After
> >>> 	./configure
> >>> 	make
> >>> 	make install
> >>>
> >>> I ran:
> >>> 	make pdf
> >>> which resulted in an error msg:
> >>> 	! LaTeX Error: File `picins.sty' not found.
> >>> which is funny, because I have a pretty complete
> >>> Latex installation (including texlive-pictures).
> >>>
> >>> I was surprised - at least in my memory I was able
> >>> to create the PDF for RC1 without a problem.
> >>>
> >>> Is this a known problem or am I doing something wrong?
> >>
> >> This seems relevant:
> >> http://sourceforge.net/p/gnuplot/patches/432/
> >
> > Thanks for pointing this out.
> >
> > I don't understand Ethan's comment to this patch:
> > I have never needed "picins" to include graphics
> > (both EPS and PDF) into Latex docs - I do it all
> > the time, without picins in my local Tex installation [...]
> 
> As I understand Ethan's comment, he's talking about flowing text 
> around an included graphic. I find I'm able to do this using
> the minipage environment plus \includegraphics from the
> graphicx package.

I see. I did not realize this.

> 
> It's not difficult to download and install picins.sty individually 
> but I agree it would be better to avoid it if possible since it's 
> not part of the current TeX Live de facto standard.

Hm, "The Latex Companion" mentions wrapfig,
which seems to provide similar functionality,
and which is part of texlive-latex-extra.

The other question is whether the wrapping 
functionality is even necessary. Just text,
followed by a centered fig, followed by more
text seems totally adequate (to me)...

But probably not worth spending a great deal
of time on (as long as people understand this
particular dependency - I did not).

Best,

		Ph.

Re: Regression problem with gp5-rc2: make pdf fails

[Allin C. <cot...@wf...>]

On Thu, 6 Nov 2014, Philipp K. Janert wrote:

> On Thu, 6 Nov 2014 20:16:30 -0500 (EST)
> Allin Cottrell <cot...@wf...> wrote:
>
>> On Thu, 6 Nov 2014, Philipp K. Janert wrote:
>>
>>>
>>> I seem to have a problem building the PDF of
>>> the doc with RC2.
>>>
>>> After
>>> 	./configure
>>> 	make
>>> 	make install
>>>
>>> I ran:
>>> 	make pdf
>>> which resulted in an error msg:
>>> 	! LaTeX Error: File `picins.sty' not found.
>>> which is funny, because I have a pretty complete
>>> Latex installation (including texlive-pictures).
>>>
>>> I was surprised - at least in my memory I was able
>>> to create the PDF for RC1 without a problem.
>>>
>>> Is this a known problem or am I doing something wrong?
>>
>> This seems relevant:
>> http://sourceforge.net/p/gnuplot/patches/432/
>
> Thanks for pointing this out.
>
> I don't understand Ethan's comment to this patch:
> I have never needed "picins" to include graphics
> (both EPS and PDF) into Latex docs - I do it all
> the time, without picins in my local Tex installation [...]

As I understand Ethan's comment, he's talking about flowing text 
around an included graphic. I find I'm able to do this using
the minipage environment plus \includegraphics from the
graphicx package.

It's not difficult to download and install picins.sty individually 
but I agree it would be better to avoid it if possible since it's 
not part of the current TeX Live de facto standard.

Allin Cottrell

Re: Regression problem with gp5-rc2: make pdf fails

[sfeam <sf...@us...>]

On Thursday, 06 November 2014 05:37:52 PM Philipp K. Janert wrote:
> On Thu, 6 Nov 2014 20:16:30 -0500 (EST)
> Allin Cottrell <cot...@wf...> wrote:
> 
> > On Thu, 6 Nov 2014, Philipp K. Janert wrote:
> > 
> > >
> > > I seem to have a problem building the PDF of
> > > the doc with RC2.
> > >
> > > After
> > > 	./configure
> > > 	make
> > > 	make install
> > >
> > > I ran:
> > > 	make pdf
> > > which resulted in an error msg:
> > > 	! LaTeX Error: File `picins.sty' not found.
> > > which is funny, because I have a pretty complete
> > > Latex installation (including texlive-pictures).
> > >
> > > I was surprised - at least in my memory I was able
> > > to create the PDF for RC1 without a problem.
> > >
> > > Is this a known problem or am I doing something wrong?
> > 
> > This seems relevant:
> > http://sourceforge.net/p/gnuplot/patches/432/
> 
> Thanks for pointing this out. 
> 
> I don't understand Ethan's comment to this patch:
> I have never needed "picins" to include graphics
> (both EPS and PDF) into Latex docs - I do it all
> the time, without picins in my local Tex installation:
> just
> 	\usepackage{graphicx}
> and
> 	\includegraphics{...}
> 
> In any case, apparently it is NOT a regression (I 
> am correcting myself here): upon further research, 
> I do not think I ever built gnuplot.pdf myself. 
> 
> Nevertheless, I wonder whether we should get rid of
> this odd dependency. 

If you can find or create an adequate replacement
for picins, feel free to contribute.
I have not found anything suitable.

	Ethan

Re: Proposal: Saving graphs from Qt and wxt terminals

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

[snip]

> 
> I've placed a patch here:
> 
> https://sourceforge.net/p/gnuplot/patches/704/
> 
> Give that a try.  All you need to do is put your code for saving as
> PNG or whatnot in the location provided in the new switch statement.
> You may also want to tweak things, e.g., change the filters to
> whatever files you can save as.
> 
> This is slightly different than the way Qt behaves as far as picking
> a file format.  It's done via the file dialog's filter feature.  It
> makes for a more conventional method of choosing, and in fact makes
> for easier programming versus putting a dropdown menu associated with
> the toolbar button.  I suggest adding your code and moving into the
> repository then let people compare/experiment.  If people prefer the
> dropdown menu, I can give that a try.

I finally got around to rebuilding from the tip
(which now includes this patch), and I think this
is fantastic! (I also like the way the format is 
chosen - just as in about every other application
out there.)

Thanks, Daniel!

I think it will really help, in particular for
casual use(r)s. 

Best,

		Ph.

Re: Regression problem with gp5-rc2: make pdf fails

[Allin C. <cot...@wf...>]

On Thu, 6 Nov 2014, Philipp K. Janert wrote:

>
> I seem to have a problem building the PDF of
> the doc with RC2.
>
> After
> 	./configure
> 	make
> 	make install
>
> I ran:
> 	make pdf
> which resulted in an error msg:
> 	! LaTeX Error: File `picins.sty' not found.
> which is funny, because I have a pretty complete
> Latex installation (including texlive-pictures).
>
> I was surprised - at least in my memory I was able
> to create the PDF for RC1 without a problem.
>
> Is this a known problem or am I doing something wrong?

This seems relevant:
http://sourceforge.net/p/gnuplot/patches/432/

Allin Cottrell

Re: Regression problem with gp5-rc2: make pdf fails

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

On Thu, 6 Nov 2014 20:16:30 -0500 (EST)
Allin Cottrell <cot...@wf...> wrote:

> On Thu, 6 Nov 2014, Philipp K. Janert wrote:
> 
> >
> > I seem to have a problem building the PDF of
> > the doc with RC2.
> >
> > After
> > 	./configure
> > 	make
> > 	make install
> >
> > I ran:
> > 	make pdf
> > which resulted in an error msg:
> > 	! LaTeX Error: File `picins.sty' not found.
> > which is funny, because I have a pretty complete
> > Latex installation (including texlive-pictures).
> >
> > I was surprised - at least in my memory I was able
> > to create the PDF for RC1 without a problem.
> >
> > Is this a known problem or am I doing something wrong?
> 
> This seems relevant:
> http://sourceforge.net/p/gnuplot/patches/432/

Thanks for pointing this out. 

I don't understand Ethan's comment to this patch:
I have never needed "picins" to include graphics
(both EPS and PDF) into Latex docs - I do it all
the time, without picins in my local Tex installation:
just
	\usepackage{graphicx}
and
	\includegraphics{...}

In any case, apparently it is NOT a regression (I 
am correcting myself here): upon further research, 
I do not think I ever built gnuplot.pdf myself. 

Nevertheless, I wonder whether we should get rid of
this odd dependency. 

Best,

		Ph.



> 
> Allin Cottrell

Regression problem with gp5-rc2: make pdf fails

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

I seem to have a problem building the PDF of 
the doc with RC2.

After
	./configure
	make
	make install

I ran:
	make pdf
which resulted in an error msg:
	! LaTeX Error: File `picins.sty' not found.
which is funny, because I have a pretty complete
Latex installation (including texlive-pictures).

I was surprised - at least in my memory I was able
to create the PDF for RC1 without a problem.

Is this a known problem or am I doing something wrong?

Best,

		Ph.

Re: NaNs not propagating

[<pl...@pi...>]

On 11/04/14 17:26, Allin Cottrell wrote:
> On Tue, 4 Nov 2014, pl...@pi... wrote:
>
> [...]
>>
>> What is the correct way to trap this condition?
>>
>> (isNaN($2))   undefined function: isNaN
>> ($2==NaN)   never true
>> ($2==0)   not true for NaN in col 2
>> (exists($2) always true, even for NaN
>>
>> Is there currently a test condition that will trap NaN in the datafile ?
>
> If it works like C, the test $2!=$2 should do it. A NaN tests as not
> equal to itself!
>
> Allin Cottrell
>

Thanks.

In principal you are correct

gnuplot >  print NaN==NaN
0


But it does not work inside plot, it's another  "not true for NaN in col 
2" case. So it seems that there is currently no way to trap this 
condition. Clearly not good.

I'll have to wait until Ethan comes up with something and update my CVS.

Peter.

Re: [Gnuplot-cvs] gnuplot ChangeLog,1.5089,1.5090

[Ethan A M. <sf...@us...>]

On Tuesday, 04 November, 2014 15:52:22 Hans-Bernhard Bröker wrote:
> Am 04.11.2014 um 05:48 schrieb Ethan Merritt:
> > Update of /cvsroot/gnuplot/gnuplot
> > In directory sfp-cvs-1.v30.ch3.sourceforge.com:/tmp/cvs-serv22050
> >
> > Modified Files:
> > 	ChangeLog
> > Log Message:
> > if $2 is NaN, plot FOO using 1:(oops=f($2)) should set oops to NaN
> 
> I have to disagree with that reasoning.  It fails to cover the case 
> where f() doesn't actually use its argument arithmetically.   E.g. if
> 
>     f(x) = (x > 0) ? (counter = counter + 1) : counter
> 
> then f(NaN) yields the current value of "counter".  It doesn't yield 
> NaN, so neither should the above "oops" receive such a value.

You are correct that the log message is a simplification of the full story.
The problem was occurring when f($2) used $2 in an arithmetic expression.
This is fixed in the new code.  As you note, logical tests are still possible.
Since logical tests against NaN will always return FALSE, it is still
easy to construct incorrect tests in the function definition.

The code change is, I think, correct.  The explanation in ChangeLog
is more complete than the 1-line commit message, but could
be expanded further to mention non-arithmetic side effects.

	Ethan

Re: NaNs not propagating

[Allin C. <cot...@wf...>]

On Tue, 4 Nov 2014, pl...@pi... wrote:

[...]
>
> What is the correct way to trap this condition?
>
> (isNaN($2))   undefined function: isNaN
> ($2==NaN)   never true
> ($2==0)   not true for NaN in col 2
> (exists($2) always true, even for NaN
>
> Is there currently a test condition that will trap NaN in the datafile ?

If it works like C, the test $2!=$2 should do it. A NaN tests as not 
equal to itself!

Allin Cottrell

Re: [Gnuplot-cvs] gnuplot ChangeLog,1.5089,1.5090

[Hans-Bernhard B. <HBB...@t-...>]

Am 04.11.2014 um 05:48 schrieb Ethan Merritt:
> Update of /cvsroot/gnuplot/gnuplot
> In directory sfp-cvs-1.v30.ch3.sourceforge.com:/tmp/cvs-serv22050
>
> Modified Files:
> 	ChangeLog
> Log Message:
> if $2 is NaN, plot FOO using 1:(oops=f($2)) should set oops to NaN

I have to disagree with that reasoning.  It fails to cover the case 
where f() doesn't actually use its argument arithmetically.   E.g. if

    f(x) = (x > 0) ? (counter = counter + 1) : counter

then f(NaN) yields the current value of "counter".  It doesn't yield 
NaN, so neither should the above "oops" receive such a value.

Re: NaNs not propagating

[<pl...@pi...>]

On 11/04/14 05:14, sfeam wrote:
> On Monday, 03 November 2014 06:45:48 PM pl...@pi... wrote:
>
>  > Hi,
>
>  >
>
>  > I seem have a problem with NaN not propagating to function calls.
>
>  >
>
>  >
>
>  > gnuplot>
>
>  >
>
>  > integ(x)=(int_last==0)?(NaN,int_last=x):(int_last=int_last+x)
>
>  >
>
>  > plot int_last=0, "-" u 1:(integ($2-44)) w l
>
>  > input data ('e' ends) > 1 1
>
>  > input data ('e' ends) > 2 2
>
>  > input data ('e' ends) > 3 NaN
>
>  > input data ('e' ends) > 4 4
>
>  > input data ('e' ends) > 5 5
>
>  > input data ('e' ends) > e
>
>  >
>
>  > gnuplot> print int_last+NaN
>
>  > NaN
>
>  >
>
>  >
>
>  > This gives a straight line with a break which means that the third line
>
>  > is evaluating NaN as zero and passing -44 to the function.
>
>  >
>
>  > I was expecting the argument (NaN-44) to pass NaN to the function which
>
>  > would in turn return and assign a NaN leaving all further points
>
>  > unplotted.
>
>  >
>
>  > Bug or feature ?
>
> I think it rates as a bug.
>
> When the data is read in, a using spec that evaluates to NaN is
>
> flagged as "undefined". This is sufficient for the purpose of
>
> plotting that one point, but as you have discovered it doesn't
>
> address side-effects of the evaluation. Arguably if the point
>
> itself is undefined then it is reasonable to say that any
>
> side-effects are also undefined, but it's less of a surprise
>
> if this makes the side-effect expressions evaluate to NaN also.
>
> I'll whip up a fix for 5.0 and 5.1.

Thanks.  I wonder what is the best way to fix this.
1. data point in undefined so fn is never called.
2. NaN propagates to fn call.

My gut feel is for 2 but I have not given it a lot of thought.


>
> By the way, in the definition you show for function integ()
>
> integ(x)=(int_last==0)?(NaN,int_last=x):(int_last=int_last+x)
>
> the NaN has no effect whatsover because it is on the left
>
> side of a comma. That means the function reduces to
>
> integ(x) = (int_last = int_last+x)
>
> Ethan
>

Thanks for pointing that out. I just lashed that up as an example but 
clearly it should be other way around  with the NaN after the assignment.

Peter.