matplotlib-devel Mailing List for matplotlib (Page 2)

On Wed, Feb 16, 2011 at 2:19 PM, Eric Firing <ef...@ha...> wrote:
> On 02/16/2011 08:38 AM, Darren Dale wrote:
>> On Wed, Feb 16, 2011 at 12:39 PM, Fernando Perez<fpe...@gm...> wrote:
>>> are you guys planning on transfering the old bugs to github?
>>
>> Probably at some point.
>>
>>> As I
>>> mentioned, I have code lying around for the upload (and to download
>>> from launchpad, but that's irrelevant here). I'm going to be mosly
>>> offline til Monday (conference trip), but if someone pings me on my
>>> Berkeley email address, which I monitor even while traveling, I'll be
>>> happy to help out.
>>
>> Thanks, that would be very helpful. I'll follow up once I figure out
>> how to extract the information from sourceforge.
>
> Darren,
>
> Just a heads-up on that: In November the tracker was heavily spammed.
> Recently I marked a few hundred items with the "delete" disposition, but
> I don't think that actually gets rid of them. If it doesn't, then maybe
> they can be filtered out during the transfer.
We have some additional spam that needs to be deleted. How did you do
it? When I try to delete several at once (mass change), I get an error
message: "XSRF Attempt Detected!"

Hi,
On Sat, Feb 26, 2011 at 6:23 AM, Pauli Virtanen <pa...@ik...> wrote:
> On 2011年2月26日 07:29:55 -0500, Michael Droettboom wrote:
> [clip]
>> Yes. The minimum version for this Python 3.x compatibility work is
>> Python 2.6. Anything earlier is just insane ;)
>
> Out of curiosity: did you consider using a single source tree for Python
> 2.x and 3.x, rather than a separate branch; providing Python 3 support by
> running 2to3 on build stage and #ifdef's + compatibility headers in C
> code?
>
> This approach allowed Numpy to support Pythons down to 2.4 with little
> extra work, and requires manual Python 3 specific fixes only when
> something breaks (which does not seem to occur often in practice).
I haven't done any Py3k porting.
I saw you (Pauli) doing the numpy and scipy ports using the same
codebase / 2to3 strategy, and that you were pleased with the results
(and it worked...).
IIRC the nose testing framework had a py3k branch that got very stale,
and they finally got there by dumping the py3k branch and using 2to3
on the same source tree.
See you,
Matthew

On 02/26/2011 04:51 AM, Darren Dale wrote:
> On Sat, Feb 26, 2011 at 9:23 AM, Pauli Virtanen<pa...@ik...> wrote:
>> On 2011年2月26日 07:29:55 -0500, Michael Droettboom wrote:
>> [clip]
>>> Yes. The minimum version for this Python 3.x compatibility work is
>>> Python 2.6. Anything earlier is just insane ;)
>>
>> Out of curiosity: did you consider using a single source tree for Python
>> 2.x and 3.x, rather than a separate branch; providing Python 3 support by
>> running 2to3 on build stage and #ifdef's + compatibility headers in C
>> code?
>>
>> This approach allowed Numpy to support Pythons down to 2.4 with little
>> extra work, and requires manual Python 3 specific fixes only when
>> something breaks (which does not seem to occur often in practice).
>
> At some point, it will make sense to discontinue support for
> <=python-2.5 (maybe even 2.6). I have used numpy's 2to3 for my
> quantities project (admittedly a *much* smaller project with *much*
> smaller exposure), but recently dropped backwards compatibility and
> now 2.6-3.2 are supported in a single branch without use of 2to3. I
> prefer this approach, it seems to be working out without too much
> trouble so far.
>
> Darren
>
Supporting earlier versions is nice--I work with many machines running 
2.5, and keep mpl up to date on several of them, but thankfully I have 
nothing still stuck on 2.4--but for mpl, now that we have a pretty good 
1.x version, I don't think it would be too hard on users if we made a 
last release supporting earlier pythons and then dropped support for 
them in subsequent releases. It would not mean people with earlier 
pythons couldn't run mpl, it would just mean they would get no new 
improvements, and no new bug fixes unless someone stepped forward to 
maintain the old branch.
Dropping 2.6 any time soon would be too drastic for my liking, though. 
It would require that any development work done on recent ubuntu 
systems, for example, would require first installing a non-default 
version of python. So, I hope the py3k work can result in a 2.6-and-up 
codebase--is that the present goal? Those of us developing on 2.6 will 
need to learn exactly how to avoid breaking 3.1 compatibility.
Eric

On Sat, Feb 26, 2011 at 8:51 AM, Darren Dale <dsd...@gm...> wrote:
> On Sat, Feb 26, 2011 at 9:23 AM, Pauli Virtanen <pa...@ik...> wrote:
> > On 2011年2月26日 07:29:55 -0500, Michael Droettboom wrote:
> > [clip]
> >> Yes. The minimum version for this Python 3.x compatibility work is
> >> Python 2.6. Anything earlier is just insane ;)
> >
> > Out of curiosity: did you consider using a single source tree for Python
> > 2.x and 3.x, rather than a separate branch; providing Python 3 support by
> > running 2to3 on build stage and #ifdef's + compatibility headers in C
> > code?
> >
> > This approach allowed Numpy to support Pythons down to 2.4 with little
> > extra work, and requires manual Python 3 specific fixes only when
> > something breaks (which does not seem to occur often in practice).
>
> At some point, it will make sense to discontinue support for
> <=python-2.5 (maybe even 2.6). I have used numpy's 2to3 for my
> quantities project (admittedly a *much* smaller project with *much*
> smaller exposure), but recently dropped backwards compatibility and
> now 2.6-3.2 are supported in a single branch without use of 2to3. I
> prefer this approach, it seems to be working out without too much
> trouble so far.
>
> Darren
>
>
My personal rule-of-thumb for what to support for scientific packages is
whichever version of python is in the latest RHEL. Python 2.6 is in RHEL6,
so I feel that is a reasonable minimum. Supporting python 2.4 (from RHEL5)
is probably too onerous at this point.
Ben Root

Hi,
On Sat, Feb 26, 2011 at 4:30 AM, Michiel de Hoon <mjl...@ya...> wrote:
>> In any case it appears that with the exception of Tkinter, it may take a
>> long time before interactive mpl backends can be used with py3k.
>
> The MacOSX backend has already been ported to Py3k (at least the C part of it, which is the largest and most difficult part of it), though I won't be able to test it until the rest of matplotlib has been ported.
>
>> The damn-ing part about the backends is that many users use only one of
>> the GUI backends, and if that one is broken for them, they believe that
>> matplotlib is completely broken. (Evidence: the user who complained
>> that matplotlib was not designed correctly when it turned out that the
>> macosx backend didn't support (non-?)interactive mode).
>
> Sometimes I wonder if it is better to retire the MacOSX backend. When it was first introduced, it was much faster (depending on what you wanted to draw) because of how its event loop is organized. By now the other backends use the same event loop strategy, and as far as I know there is no significant difference in speed compared to e.g. the TkAgg backend. The remaining advantage of the MacOSX backend is that it does not rely on other toolkits, and therefore it is easier to compile and use if something changes (e.g., when a new version of OS X comes out, or when compiling for 64-bits, or when Py3K comes out). Still, it needs some maintenance work, and it is hard to keep up.
> Opinions, anybody?
Seconding Jeff - I use the MacOSX backend too - was very grateful not
to have to install the other toolkits or suffer ugliness. So, I'm
also grateful...
Best,
Matthew

On 02/26/2011 02:30 AM, Michiel de Hoon wrote:
>> In any case it appears that with the exception of Tkinter, it may take a
>> long time before interactive mpl backends can be used with py3k.
>
> The MacOSX backend has already been ported to Py3k (at least the C part of it, which is the largest and most difficult part of it), though I won't be able to test it until the rest of matplotlib has been ported.
>
>> The damn-ing part about the backends is that many users use only one of
>> the GUI backends, and if that one is broken for them, they believe that
>> matplotlib is completely broken. (Evidence: the user who complained
>> that matplotlib was not designed correctly when it turned out that the
>> macosx backend didn't support (non-?)interactive mode).
>
> Sometimes I wonder if it is better to retire the MacOSX backend. When it was first introduced, it was much faster (depending on what you wanted to draw) because of how its event loop is organized. By now the other backends use the same event loop strategy, and as far as I know there is no significant difference in speed compared to e.g. the TkAgg backend. The remaining advantage of the MacOSX backend is that it does not rely on other toolkits, and therefore it is easier to compile and use if something changes (e.g., when a new version of OS X comes out, or when compiling for 64-bits, or when Py3K comes out). Still, it needs some maintenance work, and it is hard to keep up.
> Opinions, anybody?
Michiel,
I don't have a mac, so I can't say yes or no based on my own usage. 
Your efforts in writing and maintaining that backend are appreciated, 
though.
What about the non-interactive mode behavior--is it hard to make the 
MaxOSX backend behave like the others in that respect? From the general 
developer standpoint, I don't like divergence in behavior among the 
backends. (I don't even like the way QT4agg seems to be diverging.)
Eric
>
> --Michiel.

> Even easier is
> 
> python setupegg.py develop --user
> 
> Which doesn't require sudo and is perfect for multiple-user systems.
> I use for myself all the time.
I installed the matplotlib developement version in a virtual
environement, where you can install every package as user. Anyway,
thanks for the tip.
maximilian

On Saturday, February 26, 2011, Jouni K. Seppänen <jk...@ik...> wrote:
> Maximilian Trescher <fa...@tr...> writes:
>
>> Then i have to "install" matplotlib (with setupy.py build, setup.py
>> install). But every time i chenged something, i have to reinstall mpl to
>> test my changes.
>
> Unless you have something against setuptools (many people do):
>
> python setupegg.py develop
>
> This makes some kind of link in your site packages that points to your
> development directory, so restarting python is enough for python-level
> changes (or, presumably, reloading the relevant modules, but I have
> never understood how to do that reliably). If you change extension code,
> you'll naturally have to recompile with the same "develop" command.
>
> --
> Jouni K. Seppänen
> http://www.iki.fi/jks
>
Even easier is
python setupegg.py develop --user
Which doesn't require sudo and is perfect for multiple-user systems.
I use for myself all the time.
Ben Root
>
> ------------------------------------------------------------------------------
> Free Software Download: Index, Search & Analyze Logs and other IT data in
> Real-Time with Splunk. Collect, index and harness all the fast moving IT data
> generated by your applications, servers and devices whether physical, virtual
> or in the cloud. Deliver compliance at lower cost and gain new business
> insights. http://p.sf.net/sfu/splunk-dev2dev
> _______________________________________________
> Matplotlib-devel mailing list
> Mat...@li...
> https://lists.sourceforge.net/lists/listinfo/matplotlib-devel
>

On Sat, Feb 26, 2011 at 9:23 AM, Pauli Virtanen <pa...@ik...> wrote:
> On 2011年2月26日 07:29:55 -0500, Michael Droettboom wrote:
> [clip]
>> Yes. The minimum version for this Python 3.x compatibility work is
>> Python 2.6. Anything earlier is just insane ;)
>
> Out of curiosity: did you consider using a single source tree for Python
> 2.x and 3.x, rather than a separate branch; providing Python 3 support by
> running 2to3 on build stage and #ifdef's + compatibility headers in C
> code?
>
> This approach allowed Numpy to support Pythons down to 2.4 with little
> extra work, and requires manual Python 3 specific fixes only when
> something breaks (which does not seem to occur often in practice).
At some point, it will make sense to discontinue support for
<=python-2.5 (maybe even 2.6). I have used numpy's 2to3 for my
quantities project (admittedly a *much* smaller project with *much*
smaller exposure), but recently dropped backwards compatibility and
now 2.6-3.2 are supported in a single branch without use of 2to3. I
prefer this approach, it seems to be working out without too much
trouble so far.
Darren

On 2011年2月26日 07:29:55 -0500, Michael Droettboom wrote:
[clip]
> Yes. The minimum version for this Python 3.x compatibility work is
> Python 2.6. Anything earlier is just insane ;)
Out of curiosity: did you consider using a single source tree for Python 
2.x and 3.x, rather than a separate branch; providing Python 3 support by 
running 2to3 on build stage and #ifdef's + compatibility headers in C 
code? 
This approach allowed Numpy to support Pythons down to 2.4 with little 
extra work, and requires manual Python 3 specific fixes only when 
something breaks (which does not seem to occur often in practice).
-- 
Pauli Virtanen

On 2/26/11 5:30 AM, Michiel de Hoon wrote:
>> In any case it appears that with the exception of Tkinter, it may take a
>> long time before interactive mpl backends can be used with py3k.
> The MacOSX backend has already been ported to Py3k (at least the C part of it, which is the largest and most difficult part of it), though I won't be able to test it until the rest of matplotlib has been ported.
>
>> The damn-ing part about the backends is that many users use only one of
>> the GUI backends, and if that one is broken for them, they believe that
>> matplotlib is completely broken. (Evidence: the user who complained
>> that matplotlib was not designed correctly when it turned out that the
>> macosx backend didn't support (non-?)interactive mode).
> Sometimes I wonder if it is better to retire the MacOSX backend. When it was first introduced, it was much faster (depending on what you wanted to draw) because of how its event loop is organized. By now the other backends use the same event loop strategy, and as far as I know there is no significant difference in speed compared to e.g. the TkAgg backend. The remaining advantage of the MacOSX backend is that it does not rely on other toolkits, and therefore it is easier to compile and use if something changes (e.g., when a new version of OS X comes out, or when compiling for 64-bits, or when Py3K comes out).
Michiel
I use the macos x backend almost exclusively, for this reason. It just 
works!
It also supports more output formats than TkAgg.
-Jeff
> Still, it needs some maintenance work, and it is hard to keep up.
> Opinions, anybody?
>
> --Michiel.
>
>
>
>
>
>
> ------------------------------------------------------------------------------
> Free Software Download: Index, Search& Analyze Logs and other IT data in
> Real-Time with Splunk. Collect, index and harness all the fast moving IT data
> generated by your applications, servers and devices whether physical, virtual
> or in the cloud. Deliver compliance at lower cost and gain new business
> insights. http://p.sf.net/sfu/splunk-dev2dev
> _______________________________________________
> Matplotlib-devel mailing list
> Mat...@li...
> https://lists.sourceforge.net/lists/listinfo/matplotlib-devel

Maximilian Trescher <fa...@tr...> writes:
> Then i have to "install" matplotlib (with setupy.py build, setup.py
> install). But every time i chenged something, i have to reinstall mpl to
> test my changes.
Unless you have something against setuptools (many people do):
python setupegg.py develop
This makes some kind of link in your site packages that points to your
development directory, so restarting python is enough for python-level
changes (or, presumably, reloading the relevant modules, but I have
never understood how to do that reliably). If you change extension code,
you'll naturally have to recompile with the same "develop" command.
-- 
Jouni K. Seppänen
http://www.iki.fi/jks

It sounds like you're doing it right. Make changes to the source 
directory, compile, then test. There are some ways to streamline this. 
One is to leave a terminal window or tab open just for compiling -- then 
each compile is just a matter of "up arrow, Enter".
As an emacs user, I use the "compile" command. You can replace the 
default command that it runs ("make") to ("python setup.py install"). I 
then bind F5 to "recompile" which saves all of my unsaved buffers, and 
then reinstalls. I suspect most other programmer's editors and IDEs 
have a similar feature.
Mike
On 02/26/2011 06:31 AM, Maximilian Trescher wrote:
> Hi,
>
> i have a question about my personal developement workflow. I wanted to
> fix a bug in matplotlib, therefore i checked out the repository.
>
> Then i have to "install" matplotlib (with setupy.py build, setup.py
> install). But every time i chenged something, i have to reinstall mpl to
> test my changes.
>
> I could make my changes directly in site-packages/matplotlib, but then
> it's not version-controlled and the changes must be transferred back.
>
> This workflow seems to be suboptimal, do i miss something?
>
> Thanks,
> Maximilian
>
>
>
> ------------------------------------------------------------------------------
> Free Software Download: Index, Search& Analyze Logs and other IT data in
> Real-Time with Splunk. Collect, index and harness all the fast moving IT data
> generated by your applications, servers and devices whether physical, virtual
> or in the cloud. Deliver compliance at lower cost and gain new business
> insights. http://p.sf.net/sfu/splunk-dev2dev
>
>
> _______________________________________________
> Matplotlib-devel mailing list
> Mat...@li...
> https://lists.sourceforge.net/lists/listinfo/matplotlib-devel

On 02/25/2011 08:00 PM, Christoph Gohlke wrote:
>
> On 2/25/2011 4:03 PM, Benjamin Root wrote:
>>
>> On Fri, Feb 25, 2011 at 5:11 PM, Paul Ivanov<piv...@gm...
>> <mailto:piv...@gm...>> wrote:
>>
>> Fernando Garcia Bermudez, on 2011年02月03日 09:14, wrote:
>> > On Thu, Feb 3, 2011 at 06:05, Michael Droettboom<md...@st...
>> <mailto:md...@st...>> wrote:
>> > > On 02/03/2011 07:48 AM, Darren Dale wrote:
>> > >> On Thu, Feb 3, 2011 at 12:12 AM, Fernando Garcia Bermudez
>> > >> <mo...@gm...<mailto:mo...@gm...>> wrote:
>> > >>
>> > >>> Before continuing the debug process, though, I wanted to check
>> with
>> > >>> someone knowledgeable of this branch regarding its current
>> status. Is
>> > >>> it possible to compile matplotlib on python3?
>> > >>>
>> > >> Mike D. did some initial work for py3 support a while back. I don't
>> > >> know what platform he was working on, but he was able to produce a
>> > >> simple plot, maybe running the simple_plot demo.
>> > >>
>> > > I was working on Linux (RHEL5), and never got past that platform.
>> >
>> > I'll stay on the lookout for the github py3k branch and will do as
>> > suggested. Thanks for the update.
>> >
>>
>> By the way, there was a recent poll on python.org
>> <http://python.org> for packages
>> where users desire Python 3 support.
>>
>> Here are the top 10:
>>
>> Votes | Package
>> ------+--------
>> 837 | Django
>> 454 | wxPython
>> 406 | scipy
>> 364 | matplotlib
>> 327 | PIL
>> 266 | py2exe
>> 185 | Twisted
>> 179 | PyGTK
>> 135 | Pygame
>> 94 | web2py
>>
>> http://python.org/3kpoll
>>
>> best,
>> --
>> Paul Ivanov
>> 314 address only used for lists, off-list direct email at:
>> http://pirsquared.org | GPG/PGP key id: 0x0F3E28F7
>>
>>
>>
>> Which just goes to show that some people don't understand
>> dependencies... (PIL and PyGTK are needed by matplotllib to be fully
>> py3k compatible). Hopefully this will light a fire for them as well.
>>
>> Ben Root
>>
>
> Are PIL and PyGTK holding back matplotlib for Python 3?
PIL is used for the regression test framework, but it should be 
possible to remove that dependency there by using our own png loading code.
PyGTK is optional, of course, and users can use another GUI backend 
until the pygtk bindings catch up.
> Are there plans for matplotlib to drop Python 2.4 and 2.5? This would
> make porting to Python 3 much easier.
Yes. The minimum version for this Python 3.x compatibility work is 
Python 2.6. Anything earlier is just insane ;)
Mike
>
> Christoph
>
> ------------------------------------------------------------------------------
> Free Software Download: Index, Search& Analyze Logs and other IT data in
> Real-Time with Splunk. Collect, index and harness all the fast moving IT data
> generated by your applications, servers and devices whether physical, virtual
> or in the cloud. Deliver compliance at lower cost and gain new business
> insights. http://p.sf.net/sfu/splunk-dev2dev
> _______________________________________________
> Matplotlib-devel mailing list
> Mat...@li...
> https://lists.sourceforge.net/lists/listinfo/matplotlib-devel

> In any case it appears that with the exception of Tkinter, it may take a 
> long time before interactive mpl backends can be used with py3k.
The MacOSX backend has already been ported to Py3k (at least the C part of it, which is the largest and most difficult part of it), though I won't be able to test it until the rest of matplotlib has been ported.
> The damn-ing part about the backends is that many users use only one of
> the GUI backends, and if that one is broken for them, they believe that
> matplotlib is completely broken. (Evidence: the user who complained
> that matplotlib was not designed correctly when it turned out that the
> macosx backend didn't support (non-?)interactive mode).
Sometimes I wonder if it is better to retire the MacOSX backend. When it was first introduced, it was much faster (depending on what you wanted to draw) because of how its event loop is organized. By now the other backends use the same event loop strategy, and as far as I know there is no significant difference in speed compared to e.g. the TkAgg backend. The remaining advantage of the MacOSX backend is that it does not rely on other toolkits, and therefore it is easier to compile and use if something changes (e.g., when a new version of OS X comes out, or when compiling for 64-bits, or when Py3K comes out). Still, it needs some maintenance work, and it is hard to keep up.
Opinions, anybody?
--Michiel.
 

On 02/26/2011 02:37 AM, Eric Firing wrote:
> On 02/25/2011 04:54 PM, Benjamin Root wrote:
>>
>> On Fri, Feb 25, 2011 at 8:01 PM, John Hunter<jd...@gm...
>> <mailto:jd...@gm...>> wrote:
>>
>>
>>
>>
>>
>> On Feb 25, 2011, at 7:00 PM, Christoph Gohlke<cg...@uc...
>> <mailto:cg...@uc...>> wrote:
>>
>> > Are PIL and PyGTK holding back matplotlib for Python 3
>>
>> Not at all. Both are optional dependencies. Mpl's hard external
>> dependencies are python, zlib, libpng, freetype and numpy. With
>> those you get the agg, pdf, ps and svg backends. Various GUI
>> tooolkits are optional, as is PIL, which provides some image reading
>> capabilities and image comparisons for the unit tests.
>>
>>
>> While the GUI toolkits are optional, practically speaking, they are a
>> requirement for many (if not most) users.
>>
>> The damn-ing part about the backends is that many users use only one of
>> the GUI backends, and if that one is broken for them, they believe that
>> matplotlib is completely broken. (Evidence: the user who complained
>> that matplotlib was not designed correctly when it turned out that the
>> macosx backend didn't support (non-?)interactive mode).
>>
>> PyGTK not being updated for py3k is a heart-breaker for me. I don't
>> want to even imagine what sort of pain in the butt it is going to be to
>> program using PyGObject introspection (yeah, that sounds like fun...).
> It's not clear to me how bad it will be to handle this in the mpl
> backend. This page
> (http://live.gnome.org/PyGObject/IntrospectionPorting) suggests it might
> be quite easy, and might not require extensive code change. I think the
> idea is not so much that the bindings offer a different API but that
> their implementation is different, they are imported differently, and
> the constants are different. Unfortunately, the python-gobject package
> on Ubuntu 10.10 seems to be broken; the example that comes with it won't
> run.
>
That's right -- the changes should be minimal. The GObject 
introspection version is not meant to require the user to perform manual 
object introspection. The difference is that rather than generating 
bindings at compile-time (like the old pygtk), the method lookup and 
resolution happens at run-time (much like Python objects already work). 
The advantage is there is less of a compilation dependency between the 
version of pygtk and gtk+, and it should be much easier to get Python 
access to any library based on GObject.
> In any case it appears that with the exception of Tkinter, it may take a
> long time before interactive mpl backends can be used with py3k.
The Qt4Agg backend is also working (with the exception of the Qt form 
layout stuff) under Python 3.x.
But gtk and cairo will have to wait until the bindings settle down, fltk 
and Qt 3.x can probably be safely deprecated, and wxPython seems to be 
in Python 3.x limbo.
Mike
> Eric
>
>> Ben Root
>>
> ------------------------------------------------------------------------------
> Free Software Download: Index, Search& Analyze Logs and other IT data in
> Real-Time with Splunk. Collect, index and harness all the fast moving IT data
> generated by your applications, servers and devices whether physical, virtual
> or in the cloud. Deliver compliance at lower cost and gain new business
> insights. http://p.sf.net/sfu/splunk-dev2dev
> _______________________________________________
> Matplotlib-devel mailing list
> Mat...@li...
> https://lists.sourceforge.net/lists/listinfo/matplotlib-devel

Benjamin Root <ben...@ou...> writes:
> I am having difficulty completing a pull request that I opened. When I try
> to merge the changes to upstream, they get rejected. I can merge it back to
> my own master, and can even push it up to my github repo's master, but not
> matplotlib's master.
If you're following the merge instructions at
<http://help.github.com/pull-requests/> and getting the merge rejected,
that's likely because your master branch is not up to date. The
instructions start with "git checkout master", but that assumes your
master is the same as matplotlib's master. If that is not the case, you
should first pull from matplotlib into your master - something like
git remote add upstream gi...@gi...:matplotlib/matplotlib.git
git checkout master
git pull --ff-only upstream master
The "remote add" is necessary only once, and it gives the name
"upstream" to the matplotlib central repository (and you have probably
already done it already if you're trying to push into the repository).
Then you checkout your master branch and pull into it from upstream. The
--ff-only flag makes sure that you get an error if you have accidentally
left something on your master that is not in upstream, possibly from
your earlier merge attempts. If you do get that error, the way to clear
it is
git reset --hard upstream/master
but beware that it discards any local changes (so always commit your own
changes on feature branches, not master). (If your changes are
something that you want to save, put them on a new branch first with
"git checkout -b important-local-changes; git commit -a", and then
checkout master again and reset it to upstream/master.)
Then, once your master is up to date, you can merge the pull request and
push to upstream.
> I have tried rebasing my
> branches, but that doesn't seem to solve the problem.
Rebasing branches that you have already pushed somewhere is to be
avoided - if you rebase your feature branch, I don't think e.g. github
will know that the pull request got merged.
> I am thoroughly confused. Anybody has ideas? Should I dump my repos and
> start fresh?
No, absolutely not. This is likely just a matter of merging into an
up-to-date master branch.
-- 
Jouni K. Seppänen
http://www.iki.fi/jks

Hi,
i have a question about my personal developement workflow. I wanted to
fix a bug in matplotlib, therefore i checked out the repository.
Then i have to "install" matplotlib (with setupy.py build, setup.py
install). But every time i chenged something, i have to reinstall mpl to
test my changes.
I could make my changes directly in site-packages/matplotlib, but then
it's not version-controlled and the changes must be transferred back.
This workflow seems to be suboptimal, do i miss something?
Thanks,
Maximilian

On 02/25/2011 04:54 PM, Benjamin Root wrote:
>
>
> On Fri, Feb 25, 2011 at 8:01 PM, John Hunter <jd...@gm...
> <mailto:jd...@gm...>> wrote:
>
>
>
>
>
> On Feb 25, 2011, at 7:00 PM, Christoph Gohlke <cg...@uc...
> <mailto:cg...@uc...>> wrote:
>
> > Are PIL and PyGTK holding back matplotlib for Python 3
>
> Not at all. Both are optional dependencies. Mpl's hard external
> dependencies are python, zlib, libpng, freetype and numpy. With
> those you get the agg, pdf, ps and svg backends. Various GUI
> tooolkits are optional, as is PIL, which provides some image reading
> capabilities and image comparisons for the unit tests.
>
>
> While the GUI toolkits are optional, practically speaking, they are a
> requirement for many (if not most) users.
>
> The damn-ing part about the backends is that many users use only one of
> the GUI backends, and if that one is broken for them, they believe that
> matplotlib is completely broken. (Evidence: the user who complained
> that matplotlib was not designed correctly when it turned out that the
> macosx backend didn't support (non-?)interactive mode).
>
> PyGTK not being updated for py3k is a heart-breaker for me. I don't
> want to even imagine what sort of pain in the butt it is going to be to
> program using PyGObject introspection (yeah, that sounds like fun...).
It's not clear to me how bad it will be to handle this in the mpl 
backend. This page 
(http://live.gnome.org/PyGObject/IntrospectionPorting) suggests it might 
be quite easy, and might not require extensive code change. I think the 
idea is not so much that the bindings offer a different API but that 
their implementation is different, they are imported differently, and 
the constants are different. Unfortunately, the python-gobject package 
on Ubuntu 10.10 seems to be broken; the example that comes with it won't 
run.
In any case it appears that with the exception of Tkinter, it may take a 
long time before interactive mpl backends can be used with py3k.
Eric
>
> Ben Root
>

On Fri, Feb 25, 2011 at 10:36 PM, Christoph Gohlke <cg...@uc...> wrote:
> How up to date is matplotlib-py3 with respect to the master branch?
It is currently 5 commits behind matplotlib/master.

On 2/25/2011 6:31 PM, Darren Dale wrote:
> On Fri, Feb 25, 2011 at 9:01 PM, John Hunter<jd...@gm...> wrote:
>> On Feb 25, 2011, at 7:00 PM, Christoph Gohlke<cg...@uc...> wrote:
>>
>>> Are PIL and PyGTK holding back matplotlib for Python 3
>>
>> Not at all. Both are optional dependencies. Mpl's hard external dependencies are python, zlib, libpng, freetype and numpy. With those you get the agg, pdf, ps and svg backends. Various GUI tooolkits are optional, as is PIL, which provides some image reading capabilities and image comparisons for the unit tests.
>
> Those interested in py3 support should see
> github.com/matplotlib/matplotlib-py3 and
> https://github.com/matplotlib/matplotlib-py3/wiki . Mike D. has
> already made a good start.
>
>
Looks good. I checked out the code from git and gave it a try on 
win-amd64-py3.1. After upgrading CXX to the latest version from svn and 
making some minor modifications to the mpl code (attached), I was able 
to run many examples and also my own scripts without problem (using the 
TkAgg backend). There are more things to fix but it seems that 
matplotlib could soon be usable on Python 3.1. CXX does not yet support 
PyCapsule, which is required for Python 3.2.
How up to date is matplotlib-py3 with respect to the master branch?
Christoph

On 2011年2月25日, Darren Dale wrote:
> Here is a pull request proposing some changes to make.osx:
> https://github.com/matplotlib/matplotlib/pull/17 . I have never used
> make.osx myself. Could someone familiar with its use please have a
> look and decide whether the changes should be accepted?
>
> Thanks,
> Darren
>
Hi,
I'm not a matplotlib developer, but a user. I can certainly confirm that 
the libpng link needs to be changed. I discovered that about a month ago. 
That line in my modified version make.osx reads:
 ${PYTHON} -c 'import urllib; 
urllib.urlretrieve("http://sourceforge.net/projects/libpng/files/libpng12/older-releases/${PNGVERSION}/libpng-${PNGVERSION}.tar.gz/download", 
"libpng-${PNGVERSION}.tar.gz")' &&\
You might want to also fix the typo in:
ARCH_FLAGS="-arch i386-arch x86_64"
and then uncoment the lines that used ARCH_FLAGS, but were presumably 
commented out because of the typo and lines with "-arch i386 -arch x86_64"
added.
Cheers,
Jim

On Fri, Feb 25, 2011 at 8:01 PM, John Hunter <jd...@gm...> wrote:
>
>
>
>
> On Feb 25, 2011, at 7:00 PM, Christoph Gohlke <cg...@uc...> wrote:
>
> > Are PIL and PyGTK holding back matplotlib for Python 3
>
> Not at all. Both are optional dependencies. Mpl's hard external
> dependencies are python, zlib, libpng, freetype and numpy. With those you
> get the agg, pdf, ps and svg backends. Various GUI tooolkits are optional,
> as is PIL, which provides some image reading capabilities and image
> comparisons for the unit tests.
>
While the GUI toolkits are optional, practically speaking, they are a
requirement for many (if not most) users.
The damn-ing part about the backends is that many users use only one of the
GUI backends, and if that one is broken for them, they believe that
matplotlib is completely broken. (Evidence: the user who complained that
matplotlib was not designed correctly when it turned out that the macosx
backend didn't support (non-?)interactive mode).
PyGTK not being updated for py3k is a heart-breaker for me. I don't want to
even imagine what sort of pain in the butt it is going to be to program
using PyGObject introspection (yeah, that sounds like fun...).
Ben Root

On Fri, Feb 25, 2011 at 9:01 PM, John Hunter <jd...@gm...> wrote:
> On Feb 25, 2011, at 7:00 PM, Christoph Gohlke <cg...@uc...> wrote:
>
>> Are PIL and PyGTK holding back matplotlib for Python 3
>
> Not at all. Both are optional dependencies. Mpl's hard external dependencies are python, zlib, libpng, freetype and numpy. With those you get the agg, pdf, ps and svg backends.  Various GUI tooolkits are optional, as is PIL, which provides some image reading capabilities and image comparisons for the unit tests.
Those interested in py3 support should see
github.com/matplotlib/matplotlib-py3 and
https://github.com/matplotlib/matplotlib-py3/wiki . Mike D. has
already made a good start.

On Feb 25, 2011, at 7:00 PM, Christoph Gohlke <cg...@uc...> wrote:
> Are PIL and PyGTK holding back matplotlib for Python 3
Not at all. Both are optional dependencies. Mpl's hard external dependencies are python, zlib, libpng, freetype and numpy. With those you get the agg, pdf, ps and svg backends. Various GUI tooolkits are optional, as is PIL, which provides some image reading capabilities and image comparisons for the unit tests.