OS-X binaries?

In article
<CAH6Pt5owd7HAbZOeRZ7iTbW+Br80Uy0Z6+MpAETQemskm9VDOw@...150...>,
Matthew Brett <matthew.brett@...149...>
wrote:

Hi Chris,

> Are there recent binaries for OS-X anywhere? There don't seem to be
> any for recent releases on the MPL download page.
>
> I know we had a discussion about this a whole back, but don't remember
> the outcome. But I hope we'll continue to put them up-- macports and
> friends really aren't the best solutions for everyone.

I hope I have this cracked now, at least in principle.

The latest versions are here:

http://nipy.bic.berkeley.edu/scipy_installers/

Following Matt Terry's example, I'm testing the builds and then the
installers here:

https://travis-ci.org/matthew-brett/mpl-osx-binaries

The last ones I got from you worked very well: just a few test failures
and the current one seems to be doing about the same.

Thank you very much for providing these! I hope you will post them to
the matplotlib official site.

One odd failure (in both of them) that I don't remember seeing before:
/2.7/lib/python2.7/site-packages/matplotlib/projections/geo.py:485:
RuntimeWarning: invalid value encountered in arcsin
  theta = np.arcsin(y / np.sqrt(2))

There's a complaint about an invalid font name, but I've seen that for
quite some time:
Ekpathsea: Invalid fontname `Bitstream Vera Serif', contains ' '

FAILED (KNOWNFAIL=2, SKIP=1, errors=2)

One small suggestion: if it's not too much trouble, might you make them
.dmgs? It's a bit more convenient then having to unzip them to use them.
But if it's too much work don't bother; zipped mpkg are fine and it's
wonderful to have complete binary installers.

-- Russell

···

On Tue, Oct 22, 2013 at 9:03 AM, Chris Barker - NOAA Federal > <chris.barker@...236...> wrote:

Hi,

In article
<CAH6Pt5owd7HAbZOeRZ7iTbW+Br80Uy0Z6+MpAETQemskm9VDOw@...150...>,
Matthew Brett <matthew.brett@...149...>
wrote:

Hi Chris,

> Are there recent binaries for OS-X anywhere? There don't seem to be
> any for recent releases on the MPL download page.
>
> I know we had a discussion about this a whole back, but don't remember
> the outcome. But I hope we'll continue to put them up-- macports and
> friends really aren't the best solutions for everyone.

I hope I have this cracked now, at least in principle.

The latest versions are here:

http://nipy.bic.berkeley.edu/scipy_installers/

Following Matt Terry's example, I'm testing the builds and then the
installers here:

https://travis-ci.org/matthew-brett/mpl-osx-binaries

The last ones I got from you worked very well: just a few test failures
and the current one seems to be doing about the same.

Thank you very much for providing these! I hope you will post them to
the matplotlib official site.

I'd be happy to - I think I'm waiting for some agreement that that is
OK. I suppose I don't have permission to do that at the moment.

One odd failure (in both of them) that I don't remember seeing before:
/2.7/lib/python2.7/site-packages/matplotlib/projections/geo.py:485:
RuntimeWarning: invalid value encountered in arcsin
  theta = np.arcsin(y / np.sqrt(2))

There's a complaint about an invalid font name, but I've seen that for
quite some time:
Ekpathsea: Invalid fontname `Bitstream Vera Serif', contains ' '

FAILED (KNOWNFAIL=2, SKIP=1, errors=2)

One small suggestion: if it's not too much trouble, might you make them
.dmgs? It's a bit more convenient then having to unzip them to use them.
But if it's too much work don't bother; zipped mpkg are fine and it's
wonderful to have complete binary installers.

Yes - sure - I'll build the DMGs - was just trying to save myself some
effort while waiting for feedback - and - thanks for the feedback ...

Cheers,

Matthew

···

On Wed, Oct 23, 2013 at 11:30 AM, Russell E. Owen <rowen@...748...> wrote:

On Tue, Oct 22, 2013 at 9:03 AM, Chris Barker - NOAA Federal >> <chris.barker@...236...> wrote:

Hi,

In article
<CAH6Pt5owd7HAbZOeRZ7iTbW+Br80Uy0Z6+MpAETQemskm9VDOw@...150...>,
  Matthew Brett <matthew.brett@...149...>
  wrote:

Hi Chris,

Are there recent binaries for OS-X anywhere? There don't seem to be
any for recent releases on the MPL download page.

I know we had a discussion about this a whole back, but don't remember
the outcome. But I hope we'll continue to put them up-- macports and
friends really aren't the best solutions for everyone.

I hope I have this cracked now, at least in principle.

The latest versions are here:

http://nipy.bic.berkeley.edu/scipy_installers/

Following Matt Terry's example, I'm testing the builds and then the
installers here:

https://travis-ci.org/matthew-brett/mpl-osx-binaries

The last ones I got from you worked very well: just a few test failures
and the current one seems to be doing about the same.

Thank you very much for providing these! I hope you will post them to
the matplotlib official site.

I'd be happy to - I think I'm waiting for some agreement that that is
OK. I suppose I don't have permission to do that at the moment.

Let's talk about this at tomorrow's meeting -- or offline if you can't make the meeting. Ideally, yes, these should be posted with the other files. We can sort out the required permissions etc. offlist.

···

On 10/23/2013 02:41 PM, Matthew Brett wrote:

On Wed, Oct 23, 2013 at 11:30 AM, Russell E. Owen <rowen@...748...> wrote:

On Tue, Oct 22, 2013 at 9:03 AM, Chris Barker - NOAA Federal >>> <chris.barker@...236...> wrote:

One odd failure (in both of them) that I don't remember seeing before:
/2.7/lib/python2.7/site-packages/matplotlib/projections/geo.py:485:
RuntimeWarning: invalid value encountered in arcsin
   theta = np.arcsin(y / np.sqrt(2))

There's a complaint about an invalid font name, but I've seen that for
quite some time:
Ekpathsea: Invalid fontname `Bitstream Vera Serif', contains ' '

FAILED (KNOWNFAIL=2, SKIP=1, errors=2)

One small suggestion: if it's not too much trouble, might you make them
.dmgs? It's a bit more convenient then having to unzip them to use them.
But if it's too much work don't bother; zipped mpkg are fine and it's
wonderful to have complete binary installers.

Yes - sure - I'll build the DMGs - was just trying to save myself some
effort while waiting for feedback - and - thanks for the feedback ...

--
                    _

\/|o _|_ _. _ | | \.__ __|__|_|_ _ _ ._ _
>>(_| |(_|(/_| |_/|(_)(/_|_ |_|_)(_)(_)| | |

http://www.droettboom.com

worked well for me too (something odd with wx back end re-rendering,
but I doubt that's a Mac build issue...)

I tok a quick look at your waf scripts and I couldn't tell how you are
handling the external compiled dependencies (png, zlib, freetype) --
are these statically linked in?

It'll be good to see these posted on the MPL download site.

-Chris

···

On Wed, Oct 23, 2013 at 11:30 AM, Russell E. Owen <rowen@...748...> wrote:

The last ones I got from you worked very well: just a few test failures
and the current one seems to be doing about the same.

--

Christopher Barker, Ph.D.
Oceanographer

Emergency Response Division
NOAA/NOS/OR&R (206) 526-6959 voice
7600 Sand Point Way NE (206) 526-6329 fax
Seattle, WA 98115 (206) 526-6317 main reception

Chris.Barker@...236...

Hi,

The last ones I got from you worked very well: just a few test failures
and the current one seems to be doing about the same.

worked well for me too (something odd with wx back end re-rendering,
but I doubt that's a Mac build issue...)

I tok a quick look at your waf scripts and I couldn't tell how you are
handling the external compiled dependencies (png, zlib, freetype) --
are these statically linked in?

Yup:

https://github.com/matthew-brett/mpl-osx-binaries/blob/master/wscript#L20

through line 44 define the build rules for the libraries.

I then (this came from John H's make script I think) delete any shared
libraries:

https://github.com/matthew-brett/mpl-osx-binaries/blob/master/wscript#L183

and force mpl to link against these libraries first by setting
'basedir' in 'setup.cfg':

https://github.com/matthew-brett/mpl-osx-binaries/blob/master/wscript#L183

I should probably disable building the shared libraries as Matt T does
in his builds.

Cheers,

Matthew

···

On Wed, Oct 23, 2013 at 12:37 PM, Chris Barker <chris.barker@...236...> wrote:

On Wed, Oct 23, 2013 at 11:30 AM, Russell E. Owen <rowen@...748...> wrote: