releasing 0.6.6

[Vinzent Steinberg]

Hi all,

I'm planning to release SymPy 0.6.6 soon, hopefully before Christmas.
The last release was long ago, and many changes are accumulated. Apart
from *many* bug fixes, Aaron's ODE code (GSoC) got merged and the new
assumption system (Fabian's GSoC project) will be available but not
yet replace the old one. This is scheduled for 0.7, which will be
probably the following release. The new polys module by Mateusz should
also be merged then.
Many other things got merged, improved latex printing and improvements
to piecewise functions to name a few.

I tried to have a look at all patches that got not yet reviewed, so if
you want to see your changes in 0.6.6, please publish them for review
during the next days. Just mark them with the NeedsReview label in our
issue tracker or send them to sympy-patches.

Vinzent

[Vinzent Steinberg]

If you want to help, you can have a look at the remaining patches to
be reviewed [1] or improve existing patches [2], thanks!

Vinzent

[1] http://code.google.com/p/sympy/issues/list?q=NeedsReview
[2] http://code.google.com/p/sympy/issues/list?q=NeedsBetterPatch

[Fabian Pedregosa Izquierdo]

Vinzent Steinberg wrote:
> If you want to help, you can have a look at the remaining patches to
> be reviewed [1] or improve existing patches [2], thanks!

Thanks for managing this release, Vinzent. I'll get time to help out
when I finish my exams (next week!).

Fabian

>
> Vinzent
>
> [1] http://code.google.com/p/sympy/issues/list?q=NeedsReview
> [2] http://code.google.com/p/sympy/issues/list?q=NeedsBetterPatch
>

> --
>
> You received this message because you are subscribed to the Google Groups "sympy" group.
> To post to this group, send email to sy...@googlegroups.com.
> To unsubscribe from this group, send email to sympy+un...@googlegroups.com.
> For more options, visit this group at http://groups.google.com/group/sympy?hl=en.
>
>
>

[Aaron S. Meurer]

Also, I'm assuming we need to go through all of the Milestone-Release0.6.6 issues [1], and either fix them or postpone them. For example, I think we should see if we can fix the --random test failures (I narrowed one down in issue 1747), though some like 1244 will be best to just wait for the new assumptions in 0.7.

Let me know if you need me to review something or if you need help debugging something. I too have finals coming up, so hopefully I won't be too busy.

Aaron Meurer

[1] http://code.google.com/p/sympy/issues/list?q=label:Milestone-Release0.6.6

[Vinzent Steinberg]

On Dec 6, 6:16 pm, "Aaron S. Meurer" <asmeu...@gmail.com> wrote:
> Also, I'm assuming we need to go through all of the Milestone-Release0.6.6 issues [1], and either fix them or postpone them.  For example, I think we should see if we can fix the --random test failures (I narrowed one down in issue 1747), though some like 1244 will be best to just wait for the new assumptions in 0.7.

I think many issues which are 0.6.6 milestones are just irrelevant and
there's no one to implement it. Let's just ignore these and change the
milestone after the release. Feel free to raise the priority if you
think something should go in.

> Let me know if you need me to review something or if you need help debugging something.  I too have finals coming up, so hopefully I won't be too busy.

Currently I am trying to get integration of piecewise functions in
[1]. If you could review [2] it would be great.

Vinzent

[1] http://code.google.com/p/sympy/issues/detail?id=1695
[2] http://code.google.com/p/sympy/issues/detail?id=1748

[smichr]

I'm getting a failure on windows which is kind of weird as it is
complaining about True:

__________ sympy\queries\tests\test_query.py:test_type_extensibility
___________
File "c:\documents and settings\chris\sympy\sympy\queries\tests
\test_query.py", line 963, in test_type_extensibility
assert ask(a, Q.prime) == True
File "C:\Documents and Settings\chris\sympy\sympy\queries
\__init__.py", line 76, in ask
res = getattr(handler, subclass.__name__)(expr, assumptions)
File "<doctest queries.txt[76]>", line 4, in MyType
NameError: global name 'True' is not defined

[Ondrej Certik]

On Sun, Dec 6, 2009 at 6:35 AM, Vinzent Steinberg
<vinzent....@googlemail.com> wrote:
> If you want to help, you can have a look at the remaining patches to
> be reviewed [1] or improve existing patches [2], thanks!
>
> Vinzent
>
> [1] http://code.google.com/p/sympy/issues/list?q=NeedsReview

I reviewed all patches here.

> [2] http://code.google.com/p/sympy/issues/list?q=NeedsBetterPatch

Now we have 22 patches, that need more work. Lots of times it's some
small additions, like tests or doctests, so anyone can help with this.

I will try to improve at least my own patches soon.

Ondrej

[Vinzent Steinberg]

On Dec 8, 3:06 am, Ondrej Certik <ond...@certik.cz> wrote:
> I reviewed all patches here.
>
> > [2]http://code.google.com/p/sympy/issues/list?q=NeedsBetterPatch
>
> Now we have 22 patches, that need more work. Lots of times it's some
> small additions, like tests or doctests, so anyone can help with this.
>
> I will try to improve at least my own patches soon.

Thank you!

I updated the "changes" [1], everyone is invited to improve it. I'm
planning to add some links.

Ondrej, how did you get a list of people helping to review patches?

I'm not sure if we should sort the list by number of patches. It's not
a good criteria IMHO.

Vinzent

[1] http://code.google.com/p/sympy/wiki/Changes

[Aaron S. Meurer]

On Dec 9, 2009, at 4:45 PM, Vinzent Steinberg wrote:

> On Dec 8, 3:06 am, Ondrej Certik <ond...@certik.cz> wrote:
>> I reviewed all patches here.
>>
>>> [2]http://code.google.com/p/sympy/issues/list?q=NeedsBetterPatch
>>
>> Now we have 22 patches, that need more work. Lots of times it's some
>> small additions, like tests or doctests, so anyone can help with this.
>>
>> I will try to improve at least my own patches soon.
>
> Thank you!
>
> I updated the "changes" [1], everyone is invited to improve it. I'm
> planning to add some links.

Should I expand the part about my GSoC project to be more specific about what is implemented? How specific do we want the change log?

>
> Ondrej, how did you get a list of people helping to review patches?
>
> I'm not sure if we should sort the list by number of patches. It's not
> a good criteria IMHO.

You could do by number of lines changed (I think there is a git command for it). Or just alphabetical if you don't care about who did most.

Aaron Meurer
>
> Vinzent
>
> [1] http://code.google.com/p/sympy/wiki/Changes

[Ondrej Certik]

On Wed, Dec 9, 2009 at 3:45 PM, Vinzent Steinberg
<vinzent....@googlemail.com> wrote:
> On Dec 8, 3:06 am, Ondrej Certik <ond...@certik.cz> wrote:
>> I reviewed all patches here.
>>
>> > [2]http://code.google.com/p/sympy/issues/list?q=NeedsBetterPatch
>>
>> Now we have 22 patches, that need more work. Lots of times it's some
>> small additions, like tests or doctests, so anyone can help with this.
>>
>> I will try to improve at least my own patches soon.
>
> Thank you!
>
> I updated the "changes" [1], everyone is invited to improve it. I'm
> planning to add some links.
>
> Ondrej, how did you get a list of people helping to review patches?

I just browsed the patches and looked for signed-off-by and off top of my head.

>
> I'm not sure if we should sort the list by number of patches. It's not
> a good criteria IMHO.

Yes. Do it some better way, it's your call.

Ondrej

[Aaron S. Meurer]

There is this script to rank people by diff size (though I think it will need to be modified to only go since the sympy-0.6.5 tag).

http://git-wt-commit.rubyforge.org/git-rank-contributors

Aaron Meurer

On Dec 9, 2009, at 4:45 PM, Vinzent Steinberg wrote:

[asmeurer]

I modified it (added --since=sympy-0.6.5 to the git log command) and I
get:

dhcp-baca-10:sympy aaronmeurer(master$)$ruby ~/Downloads/git-rank-
contributors.sh -v
Vinzent Steinberg <Vinzent....@gmail.com>: 54995 lines of diff
Aaron Meurer <asme...@gmail.com>: 15582 lines of diff
Chris Smith <smi...@gmail.com>: 8117 lines of diff
Fabian Pedregosa <fab...@fseoane.net>: 7922 lines of diff
Øyvind Jensen <jensen...@gmail.com>: 3318 lines of diff
Ronan Lamy <Ronan...@normalesup.org>: 2408 lines of diff
Alan Bromborsky <abr...@verizon.net>: 2328 lines of diff
Ondrej Certik <ond...@certik.cz>: 2160 lines of diff
Toon Verstraelen <Toon.Ver...@UGent.be>: 1885 lines of diff
Jorn Baayen <jorn....@gmail.com>: 1781 lines of diff
Priit Laes <pl...@plaes.org>: 865 lines of diff
Luke Peterson <hazel...@gmail.com>: 756 lines of diff
Bill Flynn <wfl...@gmail.com>: 363 lines of diff
Akshay Srinivasan <akshaysr...@gmail.com>: 238 lines of diff
Thomas Sidoti <TSi...@gmail.com>: 220 lines of diff
Mateusz Paprocki <mat...@gmail.com>: 188 lines of diff
Fabian Seoane <fab...@fseoane.net>: 150 lines of diff
Ben Goodrich <goodri...@gmail.com>: 145 lines of diff
Ryan Krauss <ryan...@gmail.com>: 136 lines of diff
Florian Mickler <flo...@mickler.org>: 129 lines of diff
Alan Bromborsky <brombo@GA.(none)>: 122 lines of diff
Nicolas Pourcelot <nicolas....@gmail.com>: 119 lines of diff
renato <renato....@gmail.com>: 94 lines of diff
Kevin Goodsell <kevin-op...@omegacrash.net>: 89 lines of diff
Andy R. Terrel <ate...@uchicago.edu>: 78 lines of diff
Ronan Lamy <ronan...@gmail.com>: 74 lines of diff
Ted Horst <ted....@earthlink.net>: 32 lines of diff
Jochen Voss <vo...@seehuhn.de>: 24 lines of diff
Riccardo Gori <goric...@gmail.com>: 22 lines of diff
Oscar Benjamin <oscar@kar-wench.(none)>: 21 lines of diff
James Abbatiello <abb...@gmail.com>: 20 lines of diff
Fredrik Johansson <fredrik@airy.(none)>: 17 lines of diff
Eh Tan <tan2...@gmail.com>: 12 lines of diff
fabian <fabian@debian.(none)>: 8 lines of diff

Which is not too different from the number of commits ordering (given
by git shortlog --since=sympy-0.6.5 --summary --numbered).

Aaron Meurer

[Vinzent Steinberg]

On 10 déc, 02:36, asmeurer <asmeu...@gmail.com> wrote:
> I modified it (added --since=sympy-0.6.5 to the git log command) and I
> get:
>
> dhcp-baca-10:sympy aaronmeurer(master$)$ruby ~/Downloads/git-rank-
> contributors.sh -v

> Vinzent Steinberg <Vinzent.Steinb...@gmail.com>: 54995 lines of diff
> Aaron Meurer <asmeu...@gmail.com>: 15582 lines of diff

> Chris Smith <smi...@gmail.com>: 8117 lines of diff
> Fabian Pedregosa <fab...@fseoane.net>: 7922 lines of diff

> Øyvind Jensen <jensen.oyv...@gmail.com>: 3318 lines of diff
> Ronan Lamy <Ronan.L...@normalesup.org>: 2408 lines of diff
> Alan Bromborsky <abro...@verizon.net>: 2328 lines of diff

> Ondrej Certik <ond...@certik.cz>: 2160 lines of diff

> Toon Verstraelen <Toon.Verstrae...@UGent.be>: 1885 lines of diff
> Jorn Baayen <jorn.baa...@gmail.com>: 1781 lines of diff

> Priit Laes <pl...@plaes.org>: 865 lines of diff

> Luke Peterson <hazelnu...@gmail.com>: 756 lines of diff
> Bill Flynn <wfly...@gmail.com>: 363 lines of diff
> Akshay Srinivasan <akshaysriniva...@gmail.com>: 238 lines of diff
> Thomas Sidoti <TSid...@gmail.com>: 220 lines of diff
> Mateusz Paprocki <matt...@gmail.com>: 188 lines of diff

> Fabian Seoane <fab...@fseoane.net>: 150 lines of diff

> Ben Goodrich <goodrich....@gmail.com>: 145 lines of diff
> Ryan Krauss <ryanli...@gmail.com>: 136 lines of diff
> Florian Mickler <flor...@mickler.org>: 129 lines of diff

> Alan Bromborsky <brombo@GA.(none)>: 122 lines of diff

> Nicolas Pourcelot <nicolas.pource...@gmail.com>: 119 lines of diff
> renato <renato.couti...@gmail.com>: 94 lines of diff
> Kevin Goodsell <kevin-opensou...@omegacrash.net>: 89 lines of diff
> Andy R. Terrel <ater...@uchicago.edu>: 78 lines of diff
> Ronan Lamy <ronan.l...@gmail.com>: 74 lines of diff
> Ted Horst <ted.ho...@earthlink.net>: 32 lines of diff
> Jochen Voss <v...@seehuhn.de>: 24 lines of diff
> Riccardo Gori <goricca...@gmail.com>: 22 lines of diff

> Oscar Benjamin <oscar@kar-wench.(none)>: 21 lines of diff
> James Abbatiello <abb...@gmail.com>: 20 lines of diff
> Fredrik Johansson <fredrik@airy.(none)>: 17 lines of diff

> Eh Tan <tan2t...@gmail.com>: 12 lines of diff

> fabian <fabian@debian.(none)>: 8 lines of diff
>
> Which is not too different from the number of commits ordering (given
> by git shortlog --since=sympy-0.6.5 --summary --numbered).

Well, it's not just, because I just updated mpmath and pyglet,
changing lots of lines with little effort. I think I'll order it
alphabetically.

Aaron, concerning the ODE changes, I'll link to your blog. If you want
you can additionally extent your section, but it should contain only
the essential things. There is still the git changelog for extensive
documentation of changes. If you think I forgot something important or
interesting, please feel free to add it.

Vinzent

[Aaron S. Meurer]

A couple of other things:

SymPy 0.6.5 never got pushed to fink, and I don't know if it ever made it to any of the other package managers. Ondrej mentioned something about making a Debian package for it (it's mentioned here http://code.google.com/p/sympy/wiki/NewRelease).

Also, we should mention in the change log the things that are depreciated, namely, the old assumptions. All new code should use new assumptions, and old code should replace old assumptions with new, correct?

Aaron Meurer

[Ondrej Certik]

On Fri, Dec 11, 2009 at 12:41 PM, Aaron S. Meurer <asme...@gmail.com> wrote:
> A couple of other things:
>
> SymPy 0.6.5 never got pushed to fink, and I don't know if it ever made it to any of the other package managers.  Ondrej mentioned something about making a Debian package for it (it's mentioned here http://code.google.com/p/sympy/wiki/NewRelease).

Yes, I need to create the deb package (I hoped someone else can do it:).

>
> Also, we should mention in the change log the things that are depreciated, namely, the old assumptions.  All new code should use new assumptions, and old code should replace old assumptions with new, correct?

That's exactly correct.

Ondrej

[Ronan Lamy]

What exactly do you mean by this? I don't think we have any code that
uses the new assumptions and I wouldn't know how to write that since the
new assumption system isn't integrated with the old one. For instance:

In [11]: x = Symbol('x', positive=True)

In [12]: print(ask(x, Q.positive))
None

Ronan

[Ondrej Certik]

Yes, here you are using the old assumptions. I think we should
encourage to use the new ones in new modules. You are right, that we
need some code that actually uses it. I hope Fabian work on this a bit
after exams. :) But if not, it has to happen, that we slowly rewrite
things using the new assumptions.

Ondrej

[Ronan Lamy]

OK, but I am against considering the old assumptions deprecated, until
we can be sure that their replacement works as well or better.

[Ondrej Certik]

That is right. We should make this sure first.

Ondrej

[Vinzent Steinberg]

On Dec 12, 1:36 am, Ronan Lamy <ronan.l...@gmail.com> wrote:
> OK, but I am against considering the old assumptions deprecated, until
> we can be sure that their replacement works as well or better.

The new system in 0.6.6 is meant a "preview", and you can already
start using it for your new code, in some regards it much more
powerful. I removed the "deprecated" wording from the changelog.

Vinzent

[Vinzent Steinberg]

I want to release the next weekend, on December 20.

Vinzent

[Vinzent Steinberg]

I just released a release candidate, please test it [1]. I especially
don't know whether the windows installer [2] works, it would be great
if someone could try it out.

Vinzent

[1] http://sympy.googlecode.com/files/sympy-0.6.6.rc1.tar.gz
[2] http://sympy.googlecode.com/files/sympy-0.6.6.rc1.win32.exe

[David Joyner]

On Sat, Dec 19, 2009 at 8:39 PM, Vinzent Steinberg
<vinzent....@googlemail.com> wrote:
> I just released a release candidate, please test it [1]. I especially
> don't know whether the windows installer [2] works, it would be great
> if someone could try it out.
>
> Vinzent
>
> [1] http://sympy.googlecode.com/files/sympy-0.6.6.rc1.tar.gz

The source build went smoothly and as expected on a mac running 10.6.2
(python 2.6).
There seem to be some errors in the "make html" command. However, when I did
sudo cp /usr/local/bin/rst2html.py /usr/local/bin/rst2html
first, the html docs seem to build okay, except for several relatively errors.
At least, by random sampling I could not find any serious problems.
Maybe something needs to be modified in the makefile for the mac OS?

> [2] http://sympy.googlecode.com/files/sympy-0.6.6.rc1.win32.exe

[Aaron S. Meurer]

Everything including the docs works for me in Mac OS X 10.6.2. It installed fine, the isympy shortcut works in my path, and the tests all pass.

To David: I have sphinx installed through fink, which I think is a requirement for the docs to work. There are quite a few errors and warnings making the docs, but nothing that causes it to fail (it's mostly minor formatting things that Sphinx doesn't like), and nothing that we don't already know about.

The Geometric Algebra docs take forever to render, assumedly because they have LaTeX, but that is it. Do we normally ship the package with prerendered docs or just the rst files?

By the way, I noticed looking through the README that the brief history is a little outdated. Can someone who has been around since it was last updated (around 2007) bring it up to date?

Also, what are the rules for pushing in fixes during a release cycle? Should I hold off until the release, or does it not matter?

Aaron Meurer

[Vinzent Steinberg]

On Dec 20, 3:54 am, "Aaron S. Meurer" <asmeu...@gmail.com> wrote:
> Everything including the docs works for me in Mac OS X 10.6.2.  It installed fine, the isympy shortcut works in my path, and the tests all pass.  
>
> To David: I have sphinx installed through fink, which I think is a requirement for the docs to work.  There are quite a few errors and warnings making the docs, but nothing that causes it to fail (it's mostly minor formatting things that Sphinx doesn't like), and nothing that we don't already know about.  
>
> The Geometric Algebra docs take forever to render, assumedly because they have LaTeX, but that is it.  Do we normally ship the package with prerendered docs or just the rst files?

Normally we don't ship the docs, however I could release them
separately.

> By the way, I noticed looking through the README that the brief history is a little outdated.  Can someone who has been around since it was last updated (around 2007) bring it up to date?

Well, not me.

> Also, what are the rules for pushing in fixes during a release cycle?  Should I hold off until the release, or does it not matter?

I think there are no rules, feel free to push if you are sure it
introduces no regressions. I'd rather not add new features or break
the API for example.

> > The source build went smoothly and as expected on a mac running 10.6.2
> > (python 2.6).
> > There seem to be some errors in the "make html" command. However, when I did
> > sudo cp /usr/local/bin/rst2html.py /usr/local/bin/rst2html
> > first, the html docs seem to build okay, except for several relatively errors.
> > At least, by random sampling I could not find any serious problems.
> > Maybe something needs to be modified in the makefile for the mac OS?

Thanks for testing! Maybe our makefile could be improved to also
accept rst2html.py instead of rst2html.

Vinzent