ELPA and gettext.el

ELPA and gettext.el

[Alexandre Garreau]

[-- Attachment #1: Type: text/plain, Size: 348 bytes --]

Why isn’s gettext.el inside elpa? that means if you didn’t installed
gettext, or your distribution didn’t package it altogether (debian seems
to have a separate package for it), you can’t get it, or get a newer
version, with emacs, as easily as you would for the rest.

Would it be relevant and or reasonable to replicate it in ELPA?

[-- Attachment #2.1: Type: text/html, Size: 370 bytes --]

Re: ELPA and gettext.el

[Bruno Haible]

Hi,

Alexandre Garreau wrote:
> Why isn’s gettext.el inside elpa?

You mean po-mode.el?

Many years ago, it was decided to distribute po-mode.el as part of GNU gettext,
not as part of GNU Emacs, because it should also run in XEmacs.

Nowadays, it appears that XEmacs has become insignificant:
  - see https://trends.google.com/trends/explore?q=emacs,xemacs
  - in major distros it has rendering problems in the menus
  - XEmacs appears to have only 1 active developer nowadays
    https://bitbucket.org/xemacs/xemacs/commits/all

Therefore I wouldn't mind if po-mode.el was moved outside gettext, if

  1) The responsibility of po-mode.el was transferred to the Emacs community
     at the same time. Yes, it would be welcome if someone could take care of
     the 5 open bugs regarding the PO mode. [1][2][3][4][5]

  2) The documentation is handled appropriately. Currently it is documented
     through some texinfo-formatted documentation [6]. Whereas ELPA packages,
     as far as I can see, currently only have plain text documentation, which
     is inferior to texinfo / hypertext formatted documentation.

> Would it be relevant and or reasonable to replicate it in ELPA?

Or maybe into emacs/lisp/ ?

I guess that when moving it to emacs/lisp/, the documentation could be
transferred easily into the Emacs documentation. On the other hand, users
would not have the "get the newest version easily" advantage, this way.

In any case, it should be a move (transfer), not a fork.

Bruno

[1] https://savannah.gnu.org/bugs/index.php?51547
[2] https://savannah.gnu.org/bugs/index.php?55145
[3] https://savannah.gnu.org/bugs/index.php?55146
[4] https://savannah.gnu.org/bugs/index.php?55147
[5] https://savannah.gnu.org/bugs/index.php?55150
[6] https://www.gnu.org/software/gettext/manual/html_node/PO-Mode.html

Re: [bug-gettext] ELPA and gettext.el

[Paul Eggert]

Bruno Haible wrote:
> Or maybe into emacs/lisp/ ?

All things considered I think this would be the best of the proposed solutions. 
The advantage in preserving documentation outweighs the disadvantage of Emacs's 
slow release cycle. If someone would prepare a patch along those lines I can 
volunteer to review it and merge it into Emacs master.

Re: [bug-gettext] ELPA and gettext.el

[Glenn Morris]

>   1) The responsibility of po-mode.el was transferred to the Emacs community
>      at the same time. Yes, it would be welcome if someone could take care of
>      the 5 open bugs regarding the PO mode. [1][2][3][4][5]

Simply moving the files to another repository (GNU Emacs or ELPA) won't
create a maintainer for them. There is no GNU ELPA community AFAIK.
Excepting perhaps Stefan, it seems to be individuals maintaining their
own packages. Emacs itself has a small number of active people, but less
used components can languish.

>  2) The documentation is handled appropriately. Currently it is documented
>     through some texinfo-formatted documentation [6]. Whereas ELPA packages,
>     as far as I can see, currently only have plain text documentation, which
>     is inferior to texinfo / hypertext formatted documentation.

ELPA packages may contain info pages. Ref

https://www.gnu.org/software/emacs/manual/html_node/elisp/Multi_002dfile-Packages.html

Personally, I strongly recommend creating an ELPA package rather than
adding to Emacs itself.

Re: ELPA and gettext.el

[Bruno Haible]

Glenn Morris wrote:
> >  2) The documentation is handled appropriately. Currently it is documented
> >     through some texinfo-formatted documentation [6]. Whereas ELPA packages,
> >     as far as I can see, currently only have plain text documentation, which
> >     is inferior to texinfo / hypertext formatted documentation.
> 
> ELPA packages may contain info pages. Ref
> 
> https://www.gnu.org/software/emacs/manual/html_node/elisp/Multi_002dfile-Packages.html

Can ELPA packages also provide HTML-format documentation (generated from
texinfo sources) in a canonical way? HTML rendering is decent, 'info' rendering
in Emacs is inferior (use of fixed-width font for plain text, use of _xxx_
for @emph instead of slanted/italic face, etc.). Therefore I would want
to provide the documentation in HTML at least.

> Personally, I strongly recommend creating an ELPA package rather than
> adding to Emacs itself.

Hmm. I'm hearing mixed voices (Paul vs. Glenn).

> >   1) The responsibility of po-mode.el was transferred to the Emacs community
> >      at the same time. Yes, it would be welcome if someone could take care of
> >      the 5 open bugs regarding the PO mode. [1][2][3][4][5]
> 
> Simply moving the files to another repository (GNU Emacs or ELPA) won't
> create a maintainer for them. There is no GNU ELPA community AFAIK.
> Excepting perhaps Stefan, it seems to be individuals maintaining their
> own packages. Emacs itself has a small number of active people, but less
> used components can languish.

Would moving po-mode.el to emacs/lisp/ make it more probable that it gets
occasional attention and maintenance from developers familiar with Emacs Lisp,
than putting it in ELPA?

Bruno

Re: [bug-gettext] ELPA and gettext.el

[Alexandre Garreau]

[-- Attachment #1: Type: text/plain, Size: 1694 bytes --]

On 2019-01-20 at 11:58, Bruno Haible wrote:
> Can ELPA packages also provide HTML-format documentation (generated from
> texinfo sources) in a canonical way? HTML rendering is decent, 'info' rendering
> in Emacs is inferior (use of fixed-width font for plain text, use of _xxx_
> for @emph instead of slanted/italic face, etc.). Therefore I would want
> to provide the documentation in HTML at least.

A more mixed opinion about texinfo HTML output ergonomics:

As you note HTML output is here typographically superior… though, emacs
is perfectly able to use a variable-width font and italic face, probably
quite easily, but won’t (currently), afaik, be able to justify text.

Now note texinfo HTML output paragraphs are by default not justified,
and allow by default the display of way too large visual lines (longer
than the emacs 72 chars, and a looooot more than the famous “optimal”
66-character limit), and, thought that’d be only a question of generic
CSS, currently HTML output is pretty old, and absolutely not semantical
at all.  So with current HTML output you’d have to make a
custom-info-output-like plugin or browser to, from currenty,
unsemantical, HTML output, get as much semantics as the emacs info
format, which allow way better interaction and search.

A more semantical pure strict XHTML(2.0?) or HTML5 output would allow
such interaction by a decent browser (which currently doesn’t exist in
nature imho).

Now about ELPA doc: I often saw markdown (or alike, such as org-mode)
one-page documentations, which, on several package searching utilities,
automatically convert to HTML, with the typographical features you
mentioned.

[-- Attachment #2.1: Type: text/html, Size: 1909 bytes --]

Re: ELPA and gettext.el

[Paul Eggert]

On 1/20/19 2:58 AM, Bruno Haible wrote:
> Would moving po-mode.el to emacs/lisp/ make it more probable that it gets
> occasional attention and maintenance from developers familiar with Emacs Lisp,
> than putting it in ELPA?

Neither would probably give it all that much attention. I don't use ELPA 
so can't say much about it. Very occasionally people make system-wide 
sweeps of Emacs source code to catch things, and presumably this would 
include po-mode.el. However, I doubt whether there'd be much more 
attention than that unless someone filed a bug report (I hope with patches).

Re: ELPA and gettext.el

[Stefan Monnier]

> Neither would probably give it all that much attention. I don't use ELPA so
> can't say much about it. Very occasionally people make system-wide sweeps of
> Emacs source code to catch things, and presumably this would include
> po-mode.el. However, I doubt whether there'd be much more attention than
> that unless someone filed a bug report (I hope with patches).

FWIW, I do pay some (limited) attention to byte-compiler warnings, and
I do occasional `grep`s which make me bump into random pieces of code
(both in emacs.git and elpa.git).  So a change in Emacs results in
sufficiently obvious breakage in po-mode, there's a chance I'd see it
and fix it.  Pretty far from a 100% chance, tho.


        Stefan

Re: ELPA and gettext.el

[Eli Zaretskii]

> From: Stefan Monnier <monnier-CRDzTM1onBSWkKpYnGOUKg@public.gmane.org>
> Date: Fri, 25 Jan 2019 19:53:40 -0500
> Cc: bug-gettext-mXXj517/zsQ@public.gmane.org
> 
> > Neither would probably give it all that much attention. I don't use ELPA so
> > can't say much about it. Very occasionally people make system-wide sweeps of
> > Emacs source code to catch things, and presumably this would include
> > po-mode.el. However, I doubt whether there'd be much more attention than
> > that unless someone filed a bug report (I hope with patches).
> 
> FWIW, I do pay some (limited) attention to byte-compiler warnings, and
> I do occasional `grep`s which make me bump into random pieces of code
> (both in emacs.git and elpa.git).  So a change in Emacs results in
> sufficiently obvious breakage in po-mode, there's a chance I'd see it
> and fix it.  Pretty far from a 100% chance, tho.

I agree.