[bug#28000] [PATCH] gnu: octave: Build with texinfo.

[bug#28000] [PATCH] gnu: octave: Build with texinfo.

[Marius Bakke]

[-- Attachment #1: Type: text/plain, Size: 2063 bytes --]

Arun Isaac <arunisaac@systemreboot.net> writes:

> Indeed, I was waiting for a response to this patch, and am seeing your
> reply only now, after almost 2 months! It looks like I'm having trouble
> with my DNS configuration. :-(
>
>
>>> texinfo is required for the built-in documentation.
>>>
>>> * gnu/packages/maths.scm (octave)[inputs]: Add texinfo.
>>> [native-inputs]: Remove texinfo.
>>
>>If I read this right, Octave references texinfo at runtime, so it's
>>moved to inputs for correctness? The commit message is a bit confusing,
>>since it already has texinfo available :-)
>
> Currently, there is a comment above the octave package definition saying
> that users should install texinfo manually in their profile. My patch
> allows the user to not have to do that. This is the meaning I was trying
> to convey with the commit message. What other commit message would you
> suggest?

I think the comment may be referring to some of the command-line
utilities from Texinfo, e.g. for PDF exports.  There are no new
texinfo references in "octave" after this patch:

--8<---------------cut here---------------start------------->8---
$ grep -R /gnu/store/gfsiaw6rn29ii8cpjnix05a7r588k9yv-texinfo-6.3 $(./pre-inst-env guix build octave)
Binary file /gnu/store/pkg54952yx7dgv4cwjiwpfq5l9lyv90y-octave-4.2.1/lib/octave/4.2.1/liboctinterp.so.4.0.0 matches
Binary file /gnu/store/pkg54952yx7dgv4cwjiwpfq5l9lyv90y-octave-4.2.1/lib/octave/4.2.1/liboctinterp.so.4 matches
Binary file /gnu/store/pkg54952yx7dgv4cwjiwpfq5l9lyv90y-octave-4.2.1/lib/octave/4.2.1/liboctinterp.so matches
Binary file /gnu/store/pkg54952yx7dgv4cwjiwpfq5l9lyv90y-octave-4.2.1/bin/mkoctfile matches
Binary file /gnu/store/pkg54952yx7dgv4cwjiwpfq5l9lyv90y-octave-4.2.1/bin/mkoctfile-4.2.1 matches
--8<---------------cut here---------------end--------------->8---

However, native-inputs should not be referenced at all, so the patch
LGTM apart from the comment update.  Or maybe it's simply outdated?

Side note: Why is it referring to a different texinfo than the one from
`guix build texinfo`.

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 487 bytes --]

[bug#28000] [PATCH] gnu: octave: Build with texinfo.

[Arun Isaac]

Oops! I just realized that my patch doesn't work. Let me explain.

Octave uses makeinfo to render the built-in function documentation. So,
it complains when it is not able to find makeinfo. This is shown below:

octave:1> help sin
sh: makeinfo: command not found
warning: help: Texinfo formatting filter exited abnormally; raw Texinfo source of help text follows...
'sin' is a built-in function from the file libinterp/corefcn/mappers.cc

Moving texinfo to inputs does not fix this. I wrongly thought it did
because when I tested octave within the following environment, the
built-in documentation worked.

$ guix environment guix
$ ./pre-inst-env guix environment --ad-hoc octave

However, this was only because texinfo, having been pulled in by the
guix environment, was in the profile.

Now, I'll try to patch the octave source code so that it references the
makeinfo executable in the store. I will send a new patch when I'm done.

[bug#28000] [PATCH] gnu: octave: Build with texinfo.

[Arun Isaac]

texinfo is required for the built-in documentation.

* gnu/packages/maths.scm (octave)[inputs]: Add texinfo.
[native-inputs]: Remove texinfo.
---
 gnu/packages/maths.scm | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/gnu/packages/maths.scm b/gnu/packages/maths.scm
index e19c8edc8..44de8a4e6 100644
--- a/gnu/packages/maths.scm
+++ b/gnu/packages/maths.scm
@@ -1026,7 +1026,7 @@ can solve two kinds of problems:
     (license license:bsd-3)))
 
 ;; For a fully featured Octave, users  are strongly recommended also to install
-;; the following packages: texinfo, less, ghostscript, gnuplot.
+;; the following packages: less, ghostscript, gnuplot.
 (define-public octave
   (package
     (name "octave")
@@ -1058,6 +1058,7 @@ can solve two kinds of problems:
        ("glu" ,glu)
        ("zlib" ,zlib)
        ("curl" ,curl)
+       ("texinfo" ,texinfo)
        ("graphicsmagick" ,graphicsmagick)))
     (native-inputs
      `(("lzip" ,lzip)
@@ -1072,7 +1073,6 @@ can solve two kinds of problems:
        ;; will still run without them, albeit without the features they
        ;; provide.
        ("less" ,less)
-       ("texinfo" ,texinfo)
        ("ghostscript" ,ghostscript)
        ("gnuplot" ,gnuplot)))
     (arguments
-- 
2.13.4

[bug#28000] [PATCH] gnu: octave: Build with texinfo.

[Marius Bakke]

[-- Attachment #1: Type: text/plain, Size: 382 bytes --]

Arun Isaac <arunisaac@systemreboot.net> writes:

> texinfo is required for the built-in documentation.
>
> * gnu/packages/maths.scm (octave)[inputs]: Add texinfo.
> [native-inputs]: Remove texinfo.

If I read this right, Octave references texinfo at runtime, so it's
moved to inputs for correctness? The commit message is a bit confusing,
since it already has texinfo available :-)

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 487 bytes --]

[bug#28000] [PATCH] gnu: octave: Build with texinfo.

[Arun Isaac]

Indeed, I was waiting for a response to this patch, and am seeing your
reply only now, after almost 2 months! It looks like I'm having trouble
with my DNS configuration. :-(


>> texinfo is required for the built-in documentation.
>>
>> * gnu/packages/maths.scm (octave)[inputs]: Add texinfo.
>> [native-inputs]: Remove texinfo.
>
>If I read this right, Octave references texinfo at runtime, so it's
>moved to inputs for correctness? The commit message is a bit confusing,
>since it already has texinfo available :-)

Currently, there is a comment above the octave package definition saying
that users should install texinfo manually in their profile. My patch
allows the user to not have to do that. This is the meaning I was trying
to convey with the commit message. What other commit message would you
suggest?