Re: emacs-26 3fc05cf: Scripts to automate windows binary distribution

Re: emacs-26 3fc05cf: Scripts to automate windows binary distribution

[Michael Albinus]

phillip.lord@russet.org.uk (Phillip Lord) writes:

> diff --git a/admin/nt/dist-build/build-dep-zips.py b/admin/nt/dist-build/build-dep-zips.py
> new file mode 100755
> index 0000000..33bc4b5
> --- /dev/null
> +++ b/admin/nt/dist-build/build-dep-zips.py
> @@ -0,0 +1,223 @@
> +#!/usr/bin/python3

Likely I've missed the discussion, but do we want to distribute Python
scripts with Emacs?

Best regards, Michael.

Re: emacs-26 3fc05cf: Scripts to automate windows binary distribution

[Phillip Lord]

Michael Albinus <michael.albinus@gmx.de> writes:

> phillip.lord@russet.org.uk (Phillip Lord) writes:
>
>> diff --git a/admin/nt/dist-build/build-dep-zips.py
>> b/admin/nt/dist-build/build-dep-zips.py
>> new file mode 100755
>> index 0000000..33bc4b5
>> --- /dev/null
>> +++ b/admin/nt/dist-build/build-dep-zips.py
>> @@ -0,0 +1,223 @@
>> +#!/usr/bin/python3
>
> Likely I've missed the discussion, but do we want to distribute Python
> scripts with Emacs?

modules/modhelp.py has been there since 2015, so we already do.

I started off writing this in bash, but it's such a pain that I ported
it to python which ended up much nicer. Given that this file has about a
15-20 minute run time, this seemed a useful thing to me. As it happens
it was also easy to make it go quicker which is important because it
means that I can do the build in the evening, rather than over two
evenings.

It only has to run in a single environment, and that environment has
python3. I'll probably port build-zips.sh at some point as well.

Phil

Re: emacs-26 3fc05cf: Scripts to automate windows binary distribution

[Michael Albinus]

phillip.lord@russet.org.uk (Phillip Lord) writes:

Hi Phil,

>> Likely I've missed the discussion, but do we want to distribute Python
>> scripts with Emacs?
>
> modules/modhelp.py has been there since 2015, so we already do.

That's another level. It is a helper for writing external modules, and
it is not part of Emacs' infrastructure to make a distribution.

> I started off writing this in bash, but it's such a pain that I ported
> it to python which ended up much nicer. Given that this file has about a
> 15-20 minute run time, this seemed a useful thing to me. As it happens
> it was also easy to make it go quicker which is important because it
> means that I can do the build in the evening, rather than over two
> evenings.

If it were me, I would have taken Perl, I don't speak Python. Other
people would prefer Guile.

Before we continue with this, I would like to see a general
agreement. Or at least a go by the maintainers.

(Pls ignore this, if it was discussed and decided already)

The GNU Coding Standards claim:

"The standard extensibility interpreter for GNU software is Guile
(<http://www.gnu.org/software/guile/>), which implements the language
Scheme (an especially clean and simple dialect of Lisp).  [...]  We
don't reject programs written in other "scripting languages" such as
Perl and Python, but using Guile is the path that will lead to overall
consistency of the GNU system."

<https://www.gnu.org/prep/standards/standards.txt>

> Phil

Best regards, Michael.

Re: emacs-26 3fc05cf: Scripts to automate windows binary distribution

[Kaushal Modi]

[-- Attachment #1: Type: text/plain, Size: 1845 bytes --]

On Wed, Oct 25, 2017, 6:17 AM Michael Albinus <michael.albinus@gmx.de>
wrote:

>
> If it were me, I would have taken Perl, I don't speak Python. Other
> people would prefer Guile.
>
> Before we continue with this, I would like to see a general
> agreement. Or at least a go by the maintainers.
>
> (Pls ignore this, if it was discussed and decided already)
>
> The GNU Coding Standards claim:
>
> "The standard extensibility interpreter for GNU software is Guile
> (<http://www.gnu.org/software/guile/>), which implements the language
> Scheme (an especially clean and simple dialect of Lisp).  [...]  We
> don't reject programs written in other "scripting languages" such as
> Perl and Python, but using Guile is the path that will lead to overall
> consistency of the GNU system."
>
> <https://www.gnu.org/prep/standards/standards.txt>
>

Just a bystander's observation:

It would be bad if this contribution is reverted *just because* the script
is written in Python. I have gone through pain in real world where just
because the upper management doesn't like one language, stuff written in
that language has to be rewritten in another. In my experience, it's simply
a waste of time, demoralizing, and a source for introduction of new bugs.

While the "introduction of bugs" might not apply to this 1 standalone
script, it would be nice if the same bike shedding doesn't happen in the
Emacs world too.

I code in Perl, Python, Bash and other scripts. While I might have a
preference to work with one over others, it's best to keep on maintaining
the scripts written in their original languages.

If Phil is actually proficient in Guile too, then may be rewriting that
script in Guile makes some sense to support the Gnu Coding standards. But
if he is not, then I believe it will be a waste of time and loss of
motivation.

> --

Kaushal Modi

[-- Attachment #2: Type: text/html, Size: 2756 bytes --]

Re: emacs-26 3fc05cf: Scripts to automate windows binary distribution

[Noam Postavsky]

On Wed, Oct 25, 2017 at 4:37 AM, Phillip Lord
<phillip.lord@russet.org.uk> wrote:

> I started off writing this in bash, but it's such a pain that I ported
> it to python which ended up much nicer. Given that this file has about a
> 15-20 minute run time, this seemed a useful thing to me. As it happens
> it was also easy to make it go quicker which is important because it
> means that I can do the build in the evening, rather than over two
> evenings.

Hmm, looking at the python, I don't see anything that couldn't be
easily done in bash, and if it was me, I would probably use bash for
this; but if you prefer python I don't see a problem with that.

> I'll probably port build-zips.sh at some point as well.

Python seems even less advantageous there, since it's mostly calling
external commands, no?

PS: With regards to making things run quicker, you might want to
consider passing --cache-file=../../config.cache.$ARCH to configure.
It would probably save a minute or two per build.

Re: emacs-26 3fc05cf: Scripts to automate windows binary distribution

[Michael Albinus]

Kaushal Modi <kaushal.modi@gmail.com> writes:

Hi kaushal,

> It would be bad if this contribution is reverted *just because* the
> script is written in Python. I have gone through pain in real world
> where just because the upper management doesn't like one language,
> stuff written in that language has to be rewritten in another. In my
> experience, it's simply a waste of time, demoralizing, and a source
> for introduction of new bugs. 

Of course, I appreciate heavily what Phil has done.

I'm not against Python scripts per se in Emacs. I just would like to see
a decision that they are acceptable. And it shall be documented
somewhere, which kind of interpreters are accepted for Emacs scripts at all.

> While the "introduction of bugs" might not apply to this 1 standalone
> script, it would be nice if the same bike shedding doesn't happen in
> the Emacs world too. 

It's not bike shedding. It's about future maintainability of Emacs. If
everybody contributes scripts in her pet language, we will have a
situation that those scripts cannot be maintained any longer in the
future.

Best regards, Michael.

Re: emacs-26 3fc05cf: Scripts to automate windows binary distribution

[Stefan Monnier]

> Likely I've missed the discussion, but do we want to distribute Python
> scripts with Emacs?

FWIW, I don't see anything wrong with that in general.

I'd rather avoid the use of Python as a requirement for building Emacs
(not because of Python itself, but so as to avoid adding
yet-another-build-dependency), but IIUC the above script is not needed to
build Emacs, so it's OK.

> It's not bike shedding.  It's about future maintainability of Emacs.
> If everybody contributes scripts in her pet language, we will have a
> situation that those scripts cannot be maintained any longer in the
> future.

True to some extent (languages like Perl, Python, Ruby, JS, ... are
sufficiently similar to other languages that it's pretty easy to pick up
such a new language when needed, so "cannot be maintained" is a bit
pessimistic).

Tho here as well, Python is not some random pet language, so I think we
can decide that it's OK.


        Stefan

Re: emacs-26 3fc05cf: Scripts to automate windows binary distribution

[Eli Zaretskii]

> From: Michael Albinus <michael.albinus@gmx.de>
> Date: Wed, 25 Oct 2017 12:16:59 +0200
> Cc: emacs-devel@gnu.org
> 
> > I started off writing this in bash, but it's such a pain that I ported
> > it to python which ended up much nicer. Given that this file has about a
> > 15-20 minute run time, this seemed a useful thing to me. As it happens
> > it was also easy to make it go quicker which is important because it
> > means that I can do the build in the evening, rather than over two
> > evenings.
> 
> If it were me, I would have taken Perl, I don't speak Python. Other
> people would prefer Guile.
> 
> Before we continue with this, I would like to see a general
> agreement. Or at least a go by the maintainers.

FWIW, I think the preferred scripting tool in the Emacs project should
be Emacs itself.  But that's me.

I don't see a problem to have scripts in Python under admin/.  That
stuff is not for the users, it's mainly for Emacs maintainers, so the
tools it requires are less important.

Re: emacs-26 3fc05cf: Scripts to automate windows binary distribution

[Michael Albinus]

Eli Zaretskii <eliz@gnu.org> writes:

> I don't see a problem to have scripts in Python under admin/.  That
> stuff is not for the users, it's mainly for Emacs maintainers, so the
> tools it requires are less important.

Could we document, which languages are accepted there? Let's start with
Bash (or any POSIX shell), Guile, Perl, Python. Otherwise, somebody
might contribute Lua scripts, or whatever.

Best regards, Michael.

Re: emacs-26 3fc05cf: Scripts to automate windows binary distribution

[Óscar Fuentes]

Michael Albinus <michael.albinus@gmx.de> writes:

> Eli Zaretskii <eliz@gnu.org> writes:
>
>> I don't see a problem to have scripts in Python under admin/.  That
>> stuff is not for the users, it's mainly for Emacs maintainers, so the
>> tools it requires are less important.
>
> Could we document, which languages are accepted there? Let's start with
> Bash (or any POSIX shell), Guile, Perl, Python. Otherwise, somebody
> might contribute Lua scripts, or whatever.

What's wrong with Lua? :-)

IMHO, on this case, where a maintainer just puts on the repo the scripts
he wrote for his tasks, it is ok to accept the language that the
contributor decides. If those scripts where meant for users, then indeed
we should decide what's more convenient.

Re: emacs-26 3fc05cf: Scripts to automate windows binary distribution

[Eli Zaretskii]

> From: Michael Albinus <michael.albinus@gmx.de>
> Cc: phillip.lord@russet.org.uk,  emacs-devel@gnu.org
> Date: Wed, 25 Oct 2017 16:59:53 +0200
> 
> Could we document, which languages are accepted there?

Do we have to do that in advance?

> Let's start with Bash (or any POSIX shell), Guile, Perl,
> Python. Otherwise, somebody might contribute Lua scripts, or
> whatever.

If you ask me, it's the system shell and Python.  Guile is too similar
to Emacs Lisp to justify yet another scripting tool, and I don't see
why we would need Perl.

But again, I'd prefer not to start solving a problem which doesn't
seem to exist yet.  No one submitted a Lua script or even a Perl
script to us.  Let's defer debating this until it actually happens.

Re: emacs-26 3fc05cf: Scripts to automate windows binary distribution

[Michael Albinus]

Eli Zaretskii <eliz@gnu.org> writes:

> But again, I'd prefer not to start solving a problem which doesn't
> seem to exist yet.  No one submitted a Lua script or even a Perl
> script to us.  Let's defer debating this until it actually happens.

Understood, I'll shut up :-)

Best regards, Michael.

Re: emacs-26 3fc05cf: Scripts to automate windows binary distribution

[Phillip Lord]

Noam Postavsky <npostavs@users.sourceforge.net> writes:

> On Wed, Oct 25, 2017 at 4:37 AM, Phillip Lord
> <phillip.lord@russet.org.uk> wrote:
>
>> I started off writing this in bash, but it's such a pain that I ported
>> it to python which ended up much nicer. Given that this file has about a
>> 15-20 minute run time, this seemed a useful thing to me. As it happens
>> it was also easy to make it go quicker which is important because it
>> means that I can do the build in the evening, rather than over two
>> evenings.
>
> Hmm, looking at the python, I don't see anything that couldn't be
> easily done in bash, and if it was me, I would probably use bash for
> this; but if you prefer python I don't see a problem with that.

Oh, it was just the parallellisation step for the source download. It
wasn't a big deal one way or the other, and I am entirely happy to
accept that my ignorance about bash scripting is a major factor.

>> I'll probably port build-zips.sh at some point as well.
>
> Python seems even less advantageous there, since it's mostly calling
> external commands, no?

Yes, that's why I didn't port it.

> PS: With regards to making things run quicker, you might want to
> consider passing --cache-file=../../config.cache.$ARCH to configure.
> It would probably save a minute or two per build.

Good idea; for the pre-test and release builds, I do them clean anyway,
but for snapshots (assuming I start doing these), it would be a
significant win.

Phil

Re: emacs-26 3fc05cf: Scripts to automate windows binary distribution

[Phillip Lord]

Stefan Monnier <monnier@iro.umontreal.ca> writes:

>> Likely I've missed the discussion, but do we want to distribute Python
>> scripts with Emacs?
>
> FWIW, I don't see anything wrong with that in general.
>
> I'd rather avoid the use of Python as a requirement for building Emacs
> (not because of Python itself, but so as to avoid adding
> yet-another-build-dependency), but IIUC the above script is not needed to
> build Emacs, so it's OK.

This is entirely correct. It's doing the equivalent of "make-dist".

Phil

Re: emacs-26 3fc05cf: Scripts to automate windows binary distribution

[Richard Stallman]

[[[ To any NSA and FBI agents reading my email: please consider    ]]]
[[[ whether defending the US Constitution against all enemies,     ]]]
[[[ foreign or domestic, requires you to follow Snowden's example. ]]]

  > Likely I've missed the discussion, but do we want to distribute Python
  > scripts with Emacs?

I see no reason to stubbornly reject them.

All else being equal, it would be better to stick to the languages we
use now.  (Using more languages would mean more than people need to learn.)
But if there is a substantial benefit from using Python, we can.

-- 
Dr Richard Stallman
President, Free Software Foundation (gnu.org, fsf.org)
Internet Hall-of-Famer (internethalloffame.org)
Skype: No way! See stallman.org/skype.html.