gnuplot-beta

inquiry make documentation on native windows

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

In development source (5.5), 'make docs' stops at the below

*********************************************
 ! Undefined control sequence.
l.1626 \gpinsetfigure
                     {figure_E0}
?
*********************************************

All outputs are given
http://tmacchant33.starfree.jp/Files/latex-error.txt

The error does not occur in Cygwin build.

For native windows build, I have used the MikTeX that are usable  without problem until 5.4 sources.

Today I tried to use Cygwin's Texlive 2021 in the mingw build, the results are almost the same.
 
Any suggestions are welcome.

Tatsuro

Re: inquiry make documentation on native windows

[Ethan A M. <me...@uw...>]

On Saturday, 1 January 2022 21:31:47 PST Tatsuro MATSUOKA wrote:
> In development source (5.5), 'make docs' stops at the below
> 
> *********************************************
>  ! Undefined control sequence.
> l.1626 \gpinsetfigure
>                      {figure_E0}
> ?
> *********************************************
> 
> All outputs are given
> http://tmacchant33.starfree.jp/Files/latex-error.txt
> 
> The error does not occur in Cygwin build.
> 
> For native windows build, I have used the MikTeX that are usable  without problem until 5.4 sources.
> 
> Today I tried to use Cygwin's Texlive 2021 in the mingw build, the results are almost the same.
>  
> Any suggestions are welcome.


At the very top of your log file are these lines
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
   $ PATH=/f/Programs/MiKTeX/miktex/bin/x64:$PATH make docs
   cp gnuplot-figures.tex gp_tex2.tex
   echo "\usepackage{graphicx}" > pdffigures.tex
   echo "\usepackage{picins}" >> pdffigures.tex
   # Call LaTeX three times to get the toc right.
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

I think this means that your build path under Windows uses a Makefile
I did not know about and therefore failed to include when modifying
the mechanism for figure generation in commits    0a0a64   12aabd   

That would not surprise me. I broke it on linux also :-(
See Bug #2453
I tried to fix that, but it could be that I made things worse.


Is it this file?
   .../config/mingw/Makefile

How is that file generated?
Is it automatically created or is it edited by hand?

I think that the patch below would fix the problem you show,
but I don't know whether to apply it to that file directly or
apply it to a higher-level Makefile.am or Makefile.in.


%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
diff --git a/config/mingw/Makefile b/config/mingw/Makefile
index 3c578b27e..80b4824c0 100644
--- a/config/mingw/Makefile
+++ b/config/mingw/Makefile
@@ -955,6 +955,9 @@ gnuplot.pdf: gnuplot-figures.tex $(TOP)/VERSION $(D)toc_entr.sty $(D)titlepag.te
 	cp gnuplot-figures.tex gp_tex2.tex
 	echo "\usepackage{graphicx}" > pdffigures.tex
 	echo "\usepackage{picins}" >> pdffigures.tex
+	echo "\newcommand{\gpinsetfigure}[1]{" >> gpinsetfigure.tex ; \
+	echo "  \parpic[r][rt]{\includegraphics[width=3in,keepaspectratio]{#1}}" >> gpinsetfigure.tex ; \
+	echo "}" >> gpinsetfigure.tex
 	# Call LaTeX three times to get the toc right.
 	TEXINPUTS=.:$(TOP):$(D):${TEXINPUTS}: pdflatex gp_tex2.tex
 	TEXINPUTS=.:$(TOP):$(D):${TEXINPUTS}: pdflatex gp_tex2.tex
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

良いお年を

Ethan

Re: inquiry make documentation on native windows

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

Am 02.01.2022 um 08:56 schrieb Ethan A Merritt:

> Is it this file?
>     .../config/mingw/Makefile >
> How is that file generated?

It's not. All the actual makefiles in the "config" sub tree are 
hand-written, except for the source file lists they import from the 
generated src/makefile.all and src/makefile.awc.  The Makefile.am in 
there is just for organizing the tarball.

There's a case to be made that a CMake setup could replace a large part 
of those, particularly mingw, cygwin, msvc.  Among other things, that 
could reduce the extra work of maintaining them all.

I have a CMake script here that does most of what config/cygwin and 
config/msvc achieve.  I haven't yet tried to add the extra work as done 
by config/mingw: no TeX, no Japanese docs, etc.

Aw: Re: inquiry make documentation on native windows

[Bastian M. <bma...@we...>]

> Gesendet: Sonntag, 02. Januar 2022 um 14:18 Uhr
> Von: "Hans-Bernhard Bröker" <HBB...@t-...>
>
> Am 02.01.2022 um 08:56 schrieb Ethan A Merritt:
> 
> > Is it this file?
> >     .../config/mingw/Makefile >
> > How is that file generated?
> 
> It's not. All the actual makefiles in the "config" sub tree are 
> hand-written, except for the source file lists they import from the 
> generated src/makefile.all and src/makefile.awc.  The Makefile.am in 
> there is just for organizing the tarball.
> 
> There's a case to be made that a CMake setup could replace a large part 
> of those, particularly mingw, cygwin, msvc.  Among other things, that 
> could reduce the extra work of maintaining them all.

I second that. But CMake could also (mostly) replace the automake/conf 
build mechanism and hence unify the build system.
 
> I have a CMake script here that does most of what config/cygwin and 
> config/msvc achieve.  I haven't yet tried to add the extra work as done 
> by config/mingw: no TeX, no Japanese docs, etc.

I, too, have a prototype of a build system using CMake which currently 
(sort of) works for me with gcc on *nix, MinGW, MSVC, and OpenWatcom on
Windows, as well as gcc on OS/2. MacOS and Cygwin shouldn't be too hard to add.
Right now, it only covers building gnuplot and outbord terminals. It does not
cover packaging (sources, binaries, or installer), nor any documentation or
help files yet. It also still misses a number of tests.

For me, this at first started as a project to get to know CMake. I propose
to join efforts now to see if it turns out to be a viable and complete option.
Personally, I use a private github repo at the moment.
Hans-Bernhard, please let me know your thoughts on this.

  Bastian

Re: inquiry make documentation on native windows

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

Am 04.01.2022 um 08:04 schrieb Bastian Märkisch:

> I second that. But CMake could also (mostly) replace the automake/conf
> build mechanism and hence unify the build system.

IMHO, "mostly" is equivalent to "not" in this case.  As long as we still 
need some parts of the auto-foo stuff, we're better off keepin all of 
it.  It works, it does a lot for us, and covers a lot of corner cases.

I see CMake as useful only for those platforms that cannot use auto-foo,
most evidently Windows.

> For me, this at first started as a project to get to know CMake. 

Same here.

Re: Aw: Re: inquiry make documentation on native windows

[Dima K. <gn...@di...>]

> For me, this at first started as a project to get to know CMake. I
> propose to join efforts now to see if it turns out to be a viable and
> complete option. Personally, I use a private github repo at the
> moment. Hans-Bernhard, please let me know your thoughts on this.

Wait. Are we talking about redoing the whole gnuplot build system? If
so, can we please consider vanilla GNU Make? Makefiles are
straightforward to work with, and it's really nice to throw out the
cruft that cmake and autotools give you. I can put together a prototype,
if that's helpful.