Re: Python 3 support

2014-03-19 Thread Phil Holmes
- Original Message - 
From: David Kastrup d...@gnu.org

To: Julien Rioux jri...@lyx.org
Cc: LilyPond Devel lilypond-devel@gnu.org
Sent: Tuesday, March 18, 2014 2:39 PM
Subject: Re: Python 3 support


Sure.  I am not involved with GUB or the GitHub repos.  So the question
what to use for LilyPond is likely mostly up to Phil.

--
David Kastrup


I use the gperciva repo.  2.18.2 is now out and AFAIK should be the final 
2.18 build.  That's fairly irrelevant, though, since I can try updated GUB 
versions using a snapshot of the VM.  I'm in the middle of a busy time at 
college right now, but over Easter I could try some updates, if talked 
through what needs doing.


--
Phil Holmes 



___
lilypond-devel mailing list
lilypond-devel@gnu.org
https://lists.gnu.org/mailman/listinfo/lilypond-devel


Re: Python 3 support

2014-03-19 Thread David Kastrup
Phil Holmes m...@philholmes.net writes:

 - Original Message - 
 From: David Kastrup d...@gnu.org
 To: Julien Rioux jri...@lyx.org
 Cc: LilyPond Devel lilypond-devel@gnu.org
 Sent: Tuesday, March 18, 2014 2:39 PM
 Subject: Re: Python 3 support

 Sure.  I am not involved with GUB or the GitHub repos.  So the question
 what to use for LilyPond is likely mostly up to Phil.

 -- 
 David Kastrup

 I use the gperciva repo.  2.18.2 is now out and AFAIK should be the
 final 2.18 build.  That's fairly irrelevant, though, since I can try
 updated GUB versions using a snapshot of the VM.  I'm in the middle of
 a busy time at college right now, but over Easter I could try some
 updates, if talked through what needs doing.

Well, the week up to and including Easter I'll be at my yearly climbing
trip and probably not overly helpful.  But then I am not likely to be
overly helpful regarding GUB anyway.

-- 
David Kastrup

___
lilypond-devel mailing list
lilypond-devel@gnu.org
https://lists.gnu.org/mailman/listinfo/lilypond-devel


Re: Python 3 support

2014-03-18 Thread David Kastrup
Graham Percival gra...@percival-music.ca writes:

 On Sun, Mar 16, 2014 at 04:41:46PM -0400, Julien Rioux wrote:
 I think the following would be a good 1-2 punch approach for the
 current development cycle: First, upgrade python in GUB and make
 sure we can build current dev and stable branches of lilypond from
 it. Then bump the python requirement in the dev branch and start
 migrating to a codebase that supports python 2.6+ and python 3+

 Sounds great!  Thanks for working on this.

I was of the opinion that GUB already uses Python 2.6?

-- 
David Kastrup

___
lilypond-devel mailing list
lilypond-devel@gnu.org
https://lists.gnu.org/mailman/listinfo/lilypond-devel


Re: Python 3 support

2014-03-18 Thread Julien Rioux
On Tue, Mar 18, 2014 at 3:33 AM, David Kastrup d...@gnu.org wrote:

 I was of the opinion that GUB already uses Python 2.6?


GUB master at https://github.com/gperciva/gub (the current official home)
definitely does not use python 2.6. It ships python 2.4.5

If you are using GUB master at https://github.com/janneke/gub then it does
use python 2.6, but it lacks the fix for python's hashlib module, which
then fails to import at run time. I added that fix and a couple others in
the previously mentioned pull request.

(BTW moving GUB to a user-agnostic home such as
https://github.com/lilypondwould make sense to avoid such confusion.
After Jan went mostly inactive,
Graham took over as the official home, but he is now himself going into
inactivity)

Regards,
Julien
___
lilypond-devel mailing list
lilypond-devel@gnu.org
https://lists.gnu.org/mailman/listinfo/lilypond-devel


Re: Python 3 support

2014-03-18 Thread David Kastrup
Julien Rioux jri...@lyx.org writes:

 On Tue, Mar 18, 2014 at 3:33 AM, David Kastrup d...@gnu.org wrote:

 I was of the opinion that GUB already uses Python 2.6?


 GUB master at https://github.com/gperciva/gub (the current official home)
 definitely does not use python 2.6. It ships python 2.4.5

Oh.  I thought we had moved back to:

 If you are using GUB master at https://github.com/janneke/gub then it
 does use python 2.6, but it lacks the fix for python's hashlib module,
 which then fails to import at run time. I added that fix and a couple
 others in the previously mentioned pull request.

 (BTW moving GUB to a user-agnostic home such as
 https://github.com/lilypond

 would make sense to avoid such confusion.  After Jan went mostly
 inactive, Graham took over as the official home, but he is now
 himself going into inactivity)

Should we try asking Savannah, either non-GNU or GNU?  How much work
would it be to meet Savannah's licensing/guideline restrictions
regarding binary blobs and stuff?  How many of those are
LilyPond-specific?

If there are technically unavoidable obstacles, the special strategical
significance of GUB might still make it possible to negotiate about the
hosting with Richard Stallman, currently the ultimate decision maker.

I think that cross-platform support is currently troublesome for enough
projects that the compile under GNU/Linux, provide everywhere approach
of GUB would mean a significant concentration of efforts for other GNU
projects as well.

-- 
David Kastrup

___
lilypond-devel mailing list
lilypond-devel@gnu.org
https://lists.gnu.org/mailman/listinfo/lilypond-devel


Re: Python 3 support

2014-03-18 Thread Julien Rioux
On Tue, Mar 18, 2014 at 6:24 AM, David Kastrup d...@gnu.org wrote:

 Julien Rioux jri...@lyx.org writes:

  (BTW moving GUB to a user-agnostic home such as
  https://github.com/lilypond

  would make sense to avoid such confusion.  After Jan went mostly
  inactive, Graham took over as the official home, but he is now
  himself going into inactivity)

 Should we try asking Savannah, either non-GNU or GNU?  How much work
 would it be to meet Savannah's licensing/guideline restrictions
 regarding binary blobs and stuff?  How many of those are
 LilyPond-specific?

 If there are technically unavoidable obstacles, the special strategical
 significance of GUB might still make it possible to negotiate about the
 hosting with Richard Stallman, currently the ultimate decision maker.

 I think that cross-platform support is currently troublesome for enough
 projects that the compile under GNU/Linux, provide everywhere approach
 of GUB would mean a significant concentration of efforts for other GNU
 projects as well.


If you are keen on it, why not? Not sure if it's worth the trouble, though:
Maybe more visibility would bring GUB more workers, and in that vein
endorsement by a big player would be a boost. Unfortunately, I'm not sure
GUB has a strong significance anymore. With no maintainer, no support team,
virtually no devs, it's not attractive for new projects, which then opt for
other cross-build solutions instead.

The current hosting situation isn't bad that we need to take such important
actions with savannah. With github, we already have hosting, a platform for
contribution and review comments, and relatively strong visibility, at no
effort from us. It seems there is confusion about who owns the official
repo, which is easily solved if we move the repo from an individual to an
organization. And since it's literally a click or two to do that, I though
I would suggest it in passing. Anyway, we're sidestepping.

Regards,
Julien
___
lilypond-devel mailing list
lilypond-devel@gnu.org
https://lists.gnu.org/mailman/listinfo/lilypond-devel


Re: Python 3 support

2014-03-18 Thread Julien Rioux
On Mon, Mar 17, 2014 at 9:38 PM, Graham Percival
gra...@percival-music.cawrote:

 On Sun, Mar 16, 2014 at 04:41:46PM -0400, Julien Rioux wrote:
  I think the following would be a good 1-2 punch approach for the
  current development cycle: First, upgrade python in GUB and make
  sure we can build current dev and stable branches of lilypond from
  it. Then bump the python requirement in the dev branch and start
  migrating to a codebase that supports python 2.6+ and python 3+

 Sounds great!  Thanks for working on this.


When would be a good timing to do the switch in GUB? Probably don't want to
mess with releases from the current stable (2.18.x) branch at this point,
so maybe wait until after the last 2.18.x is released, whenever that might
be?

Regards,
Julien
___
lilypond-devel mailing list
lilypond-devel@gnu.org
https://lists.gnu.org/mailman/listinfo/lilypond-devel


Re: Python 3 support

2014-03-18 Thread David Kastrup
Julien Rioux jri...@lyx.org writes:

 If you are keen on it, why not? Not sure if it's worth the trouble,
 though: Maybe more visibility would bring GUB more workers, and in
 that vein endorsement by a big player would be a boost. Unfortunately,
 I'm not sure GUB has a strong significance anymore. With no
 maintainer, no support team, virtually no devs, it's not attractive
 for new projects, which then opt for other cross-build solutions
 instead.

There is not a lot around to opt for as far as I know.

 The current hosting situation isn't bad that we need to take such
 important actions with savannah. With github, we already have hosting,
 a platform for contribution and review comments, and relatively strong
 visibility, at no effort from us.

Who is we?  I for one am not going to agree to GitHub's quite invasive
usage conditions for their free offerings which include killing a
project for any reason they want to at any point of time, explicitly
citing bandwidth usage as one such reason.

Now the situation is in theory not all that different with Savannah.
Except that Savannah does not serve stockholders but its users and Free
Software.  When they find that there is a technical problem in
connection with serving a project, pull the plug will be way lower in
the list of remedies than with GitHub.

 It seems there is confusion about who owns the official repo, which
 is easily solved if we move the repo from an individual to an
 organization. And since it's literally a click or two to do that, I
 though I would suggest it in passing. Anyway, we're sidestepping.

Sure.  I am not involved with GUB or the GitHub repos.  So the question
what to use for LilyPond is likely mostly up to Phil.

-- 
David Kastrup

___
lilypond-devel mailing list
lilypond-devel@gnu.org
https://lists.gnu.org/mailman/listinfo/lilypond-devel


Re: Python 3 support

2014-03-18 Thread Julien Rioux
On Tue, Mar 18, 2014 at 10:39 AM, David Kastrup d...@gnu.org wrote:

 Julien Rioux jri...@lyx.org writes:

  The current hosting situation isn't bad that we need to take such
  important actions with savannah. With github, we already have hosting,
  a platform for contribution and review comments, and relatively strong
  visibility, at no effort from us.

 Who is we?  I for one am not going to agree to GitHub's quite invasive
 usage conditions for their free offerings which include killing a
 project for any reason they want to at any point of time, explicitly
 citing bandwidth usage as one such reason.


You're right, for contributing a patch and/or commenting through github,
it's we as in those that are bold enough to sign up on github. We
also happily send patches or comments on the mailing list as usual; we as
in those that are bold enough to post on public mailing lists. It's still
an added value that a non-empty set of people are happy with.


 Now the situation is in theory not all that different with Savannah.
 Except that Savannah does not serve stockholders but its users and Free
 Software.  When they find that there is a technical problem in
 connection with serving a project, pull the plug will be way lower in
 the list of remedies than with GitHub.


Fortunately with git repos even if the hosting goes, the project's source
code is still in hand,

Regards,
Julien
___
lilypond-devel mailing list
lilypond-devel@gnu.org
https://lists.gnu.org/mailman/listinfo/lilypond-devel


Re: Python 3 support

2014-03-18 Thread David Kastrup
Julien Rioux jri...@lyx.org writes:

 On Tue, Mar 18, 2014 at 10:39 AM, David Kastrup d...@gnu.org wrote:

 Julien Rioux jri...@lyx.org writes:

  The current hosting situation isn't bad that we need to take such
  important actions with savannah. With github, we already have hosting,
  a platform for contribution and review comments, and relatively strong
  visibility, at no effort from us.

 Who is we?  I for one am not going to agree to GitHub's quite invasive
 usage conditions for their free offerings which include killing a
 project for any reason they want to at any point of time, explicitly
 citing bandwidth usage as one such reason.


 You're right, for contributing a patch and/or commenting through github,
 it's we as in those that are bold enough to sign up on github. We
 also happily send patches or comments on the mailing list as usual; we as
 in those that are bold enough to post on public mailing lists. It's still
 an added value that a non-empty set of people are happy with.


 Now the situation is in theory not all that different with Savannah.
 Except that Savannah does not serve stockholders but its users and Free
 Software.  When they find that there is a technical problem in
 connection with serving a project, pull the plug will be way lower in
 the list of remedies than with GitHub.


 Fortunately with git repos even if the hosting goes, the project's source
 code is still in hand,

That's not quite the same as we already have hosting, a platform for
contribution and review comments.  Everything beyond the content in
private repositories is gone if a project is removed.  And we have is
a bit of a euphemism for proprietary software run on a proprietary
service with a proprietary data store for everything but the central
repository itself.  It's more like we are permitted to use.

Of course, with Savannah we have the same situation regarding the we
are permitted to use angle, but the motivations for the permission are
different.  That makes me feel more at home.

-- 
David Kastrup

___
lilypond-devel mailing list
lilypond-devel@gnu.org
https://lists.gnu.org/mailman/listinfo/lilypond-devel


Re: Python 3 support

2014-03-18 Thread Julien Rioux
On Tue, Mar 18, 2014 at 12:44 PM, David Kastrup d...@gnu.org wrote:


 That's not quite the same as we already have hosting, a platform for
 contribution and review comments.  Everything beyond the content in
 private repositories is gone if a project is removed.  And we have is
 a bit of a euphemism for proprietary software run on a proprietary
 service with a proprietary data store for everything but the central
 repository itself.  It's more like we are permitted to use.

 Of course, with Savannah we have the same situation regarding the we
 are permitted to use angle, but the motivations for the permission are
 different.  That makes me feel more at home.


Again, if you're keen on providing hosting elsewhere, why not go for it. My
point of view says it is not worth it, yours differ. I'll be glad if we get
back on thread.

Regards,
Julien
___
lilypond-devel mailing list
lilypond-devel@gnu.org
https://lists.gnu.org/mailman/listinfo/lilypond-devel


Re: Python 3 support

2014-03-18 Thread David Kastrup
Julien Rioux jri...@lyx.org writes:

 On Tue, Mar 18, 2014 at 12:44 PM, David Kastrup d...@gnu.org wrote:


 That's not quite the same as we already have hosting, a platform for
 contribution and review comments.  Everything beyond the content in
 private repositories is gone if a project is removed.  And we have is
 a bit of a euphemism for proprietary software run on a proprietary
 service with a proprietary data store for everything but the central
 repository itself.  It's more like we are permitted to use.

 Of course, with Savannah we have the same situation regarding the we
 are permitted to use angle, but the motivations for the permission are
 different.  That makes me feel more at home.


 Again, if you're keen on providing hosting elsewhere, why not go for
 it. My point of view says it is not worth it, yours differ.

I'm not likely to invest myself much with GUB myself.  The main
incentive would be to do this as part of trying to align oneself closer
with the GNU project in the hope/expectations that others use and
contribute to GUB as a consequence.

How much the LilyPond project would be profiting as a result if at all
is not something I can really estimate.  It is clear that at the current
point of time, not much is happening with GUB, but it's not all that
clear that other people's work on GUB would in turn help LilyPond's use
case.

 I'll be glad if we get back on thread.

Well, you were talking about a github-internal move.  First step would
likely be to communicate with Phil and figure out what GUB repository he
is actually working with.  His internet provider apparently is currently
Ddos-attacked, so we probably need to wait until this calms down.

-- 
David Kastrup

___
lilypond-devel mailing list
lilypond-devel@gnu.org
https://lists.gnu.org/mailman/listinfo/lilypond-devel


Re: Python 3 support

2014-03-18 Thread Jeremiah Benham
I have forked gub and have been working on it for a while now.

https://github.com/jjbenham/gub

It is very different now from https://github.com/gperciva/gub and the main
master. I don't know how to contribute changes unless via per file basis.
Then each patch would need to be tested. I have added support for gtk3 on
mingw, darwin-x86 and linux-x86. I also upgraded many things like tar (this
adds .xz support). This was all to support the denemo project. It would be
nice to work with others on it.

Jeremiah
On Mar 18, 2014 4:58 AM, Julien Rioux jri...@lyx.org wrote:

 On Tue, Mar 18, 2014 at 3:33 AM, David Kastrup d...@gnu.org wrote:

 I was of the opinion that GUB already uses Python 2.6?


 GUB master at https://github.com/gperciva/gub (the current official
 home) definitely does not use python 2.6. It ships python 2.4.5

 If you are using GUB master at https://github.com/janneke/gub then it
 does use python 2.6, but it lacks the fix for python's hashlib module,
 which then fails to import at run time. I added that fix and a couple
 others in the previously mentioned pull request.

 (BTW moving GUB to a user-agnostic home such as
 https://github.com/lilypond would make sense to avoid such confusion.
 After Jan went mostly inactive, Graham took over as the official home,
 but he is now himself going into inactivity)

 Regards,
 Julien

 ___
 lilypond-devel mailing list
 lilypond-devel@gnu.org
 https://lists.gnu.org/mailman/listinfo/lilypond-devel


___
lilypond-devel mailing list
lilypond-devel@gnu.org
https://lists.gnu.org/mailman/listinfo/lilypond-devel


Re: Python 3 support

2014-03-18 Thread Julien Rioux
On Tue, Mar 18, 2014 at 4:43 PM, Jeremiah Benham jjben...@chicagoguitar.com
 wrote:

 I have forked gub and have been working on it for a while now.

 https://github.com/jjbenham/gub

 It is very different now from https://github.com/gperciva/gub and the
 main master. I don't know how to contribute changes unless via per file
 basis. Then each patch would need to be tested. I have added support for
 gtk3 on mingw, darwin-x86 and linux-x86. I also upgraded many things like
 tar (this adds .xz support). This was all to support the denemo project. It
 would be nice to work with others on it.

 Jeremiah


Hi Jeremiah,

Thanks for reaching out. A while ago I noticed your fork and though that at
some point it would be good to investigate what should be contributed back.
Unfortunately it takes a lot of time to test changes in GUB, and I was
already puzzled enough by the various different branches between Jan's and
Graham's repos. I'm glad to hear that you would like to contribute your
changes back, that will be a welcome collaboration.

Cheers,
Julien
___
lilypond-devel mailing list
lilypond-devel@gnu.org
https://lists.gnu.org/mailman/listinfo/lilypond-devel


Re: Python 3 support

2014-03-17 Thread Graham Percival
On Sun, Mar 16, 2014 at 04:41:46PM -0400, Julien Rioux wrote:
 I think the following would be a good 1-2 punch approach for the
 current development cycle: First, upgrade python in GUB and make
 sure we can build current dev and stable branches of lilypond from
 it. Then bump the python requirement in the dev branch and start
 migrating to a codebase that supports python 2.6+ and python 3+

Sounds great!  Thanks for working on this.

Cheers,
- Graham

___
lilypond-devel mailing list
lilypond-devel@gnu.org
https://lists.gnu.org/mailman/listinfo/lilypond-devel


Python 3 support

2014-03-16 Thread Julien Rioux

Hi,

Our python scripts currently require a python2 interpreter with version 
at least 2.4, but by bumping the requirement to 2.6 we could start 
introducing

from __future__ import print_function
which will be one of the major step towards supporting both python2 and 
python3 interpreters with a single codebase.


One of the big hurdle to bumping the python version requirement is our 
own cross-platform build system GUB which provides 2.4, but last summer 
I was able to build lilypond with an updated GUB, based on some 
experimental branch by Jan where I added a few commits. You can find the 
patchset here: https://github.com/gperciva/gub/pull/6


I think the following would be a good 1-2 punch approach for the current 
development cycle: First, upgrade python in GUB and make sure we can 
build current dev and stable branches of lilypond from it. Then bump the 
python requirement in the dev branch and start migrating to a codebase 
that supports python 2.6+ and python 3+


Thoughts? Objections?

Regards,
Julien


___
lilypond-devel mailing list
lilypond-devel@gnu.org
https://lists.gnu.org/mailman/listinfo/lilypond-devel