matplotlib-users Mailing List for matplotlib

It may be a minor one, but can the web manager delete the last sentence on
this page? It is the same as the previous sentence...

http://matplotlib.sourceforge.net/devel/coding_guide.html

~Pengkui

Hi I am trying to create an executable for a wxpython app that plots
some data using matplotlib , pyplot.figure().

I have tried following the directions for including the matplotlib
data files detailed here(http://www.py2exe.org/index.cgi/MatPlotLib)
but still get an error indicating that the data files are not bundled.


RuntimeError: Could not find matplotlib data files



#######################
Here is my setup_win.py
#######################

from distutils.core import setup
import py2exe
import matplotlib


from distutils.filelist import findall
import os

name = "CuppaTmGui.py"
DATA_FILES = matplotlib.get_py2exe_datafiles()
#DATA_FILES.append(r"C:\Python27_32\tcl\tcl8.5\init.tcl")
setup(console=['CuppaTmGui.py'], options = { "py2exe" : {"includes" :
["cuppa_tm","matplotlib.backends.backend_tkagg","numpy","pylab"],"skip_archive":0
, "bundle_files" : 1 }}, data_files = DATA_FILES)



I am wondering what I am doing wrong while packaging the executable.

I  also tried to run with "bundle_files : 2"  option or "bundle_files"
: 3  , but that results in an exe that immediately crashes and
triggers a miscrosoft error report.
I am building my exe using 32 bit python on a 64 bit windows machine.


Thanks for your help

Hari

Dear Michael,

Sorry for such a lengthy email, but I wanted to ask about dependencies
as well. I am trying to get python and py3k versions of matplotlib
working. In addition to this I am trying to do this with git/master
version. I can build Python2 version, but I could not do it with Python3.

Could you tell me which packages are indeed needed to build python
git/master? What are differences between Python2 and Python3
dependencies? I could later provide a patch for the git/master INSTALL
file.

So, the compile dependencies are as follows?
    media-libs/libpng   # Needed?
    dev-python/imaging  # Only for docs?
    dev-python/xlwt     # What about this package?

Runtime dependencies?
    virtual/ttf-fonts       # I guess theese are needed for selecting
    media-fonts/stix-fonts  # different fonts, or is it the case that
    media-fonts/texcm-ttf   # mpl uses LaTeX fonts already?
    dev-python/pyparsing    # Necessary?
    cairo?  ( dev-python/pycairo ) # What about pango?
    excel?  ( dev-python/xlwt )
    latex?  (               # Needed only for latex'ing
        virtual/latex-base
        app-text/ghostscript-gpl
        app-text/poppler[utils] # Needed?
        dev-texlive/texlive-fontsrecommended
    )"

I started to follow the development of MPL a bit and IIRC it is
now possible to install a python3 version along with the Python2. What
is more one can use Pyside instead of pyqt4 for qt. But what about gtk?
What libs are needed if I want to use gtk and python3? I remember, that
there were some issues, but do not know whether they were resolved.

Maybe it would be a good idea to reflect the Python3 status/dependencies
in the documentation?

Either way, thanks very much for working on the project!

All best,
Ignas A.

See PR #855

https://github.com/matplotlib/matplotlib/pull/855

Mike

On 05/03/2012 10:01 AM, Michael Droettboom wrote:
> On 05/03/2012 09:48 AM, Benjamin Root wrote:
>>
>>
>> On Wed, May 2, 2012 at 3:59 PM, David Erickson <der...@st... 
>> <mailto:der...@st...>> wrote:
>>
>>     On 5/1/2012 5:41 AM, Benjamin Root wrote:
>>>
>>>
>>>     On Tue, May 1, 2012 at 6:42 AM, David Erickson
>>>     <der...@st... <mailto:der...@st...>> wrote:
>>>
>>>         Hi I need to use the symlog yscale in my graph, I have a lot
>>>         of data
>>>         that needs to be displayed linearly, with a small fraction
>>>         at the upper
>>>         end of the range that needs to be displayed in log scale due
>>>         to its
>>>         distance from the main data.  The symlog scale works great,
>>>         however I'd
>>>         like to adjust the actual vertical graph position where it
>>>         crosses from
>>>         linear to log (not the threshold), because right now only
>>>         ~25% of the y
>>>         space is being given to linear, and I'd like it to be more
>>>         like 80%.  Is
>>>         this possible?
>>>
>>>         Thanks!
>>>         David
>>>
>>>
>>>     Isn't linthreshy what you are looking for?  It denotes the range
>>>     where the scale is linear.  So, if it is 25, then from -25 to 25
>>>     the scale will be linear.  After 25, it will be log.
>>>
>>>     Maybe I am missing something in your description?
>>>
>>>     Cheers!
>>>     Ben Root
>>>
>>
>>     Hi Ben,
>>     No unfortunately linthreshy only controls the crossover point
>>     from linear to log scale, it does not give you control over where
>>     this occurs on the figure's y coordinates.  I've attached a
>>     picture to explain, currently the linear part of the graph is
>>     only being given around 25% of the vertical space, I'd like to
>>     reverse that and give linear ~75% and log at the top only the
>>     remaining 25%.  Is this possible?  I've been digging around in
>>     the SymmetricalLogScale and SymmetricalLogLocator classes and
>>     can't even tell how this range is allocated.
>>
>>     Thanks in advance!!
>>     -David
>>
>>
>> David,
>>
>> Thanks, that is much clearer what you are looking for.  You are 
>> right, I can't seem to find any sort of obvious way to get what you 
>> want.  Reading the docs for the SymLog scale indicates to me that the 
>> author intended for the logrithmic portion to be most interesting and 
>> the linear portion was only supposed to be a work-around the whole 
>> log(0) issue.  Could you file an issue on the github page so that we 
>> can mark it as a feature request?
>
> Just getting to this after some unexpected absences.  Yes -- that is 
> how symlog was originally intended, so there isn't currently any way 
> to configure it.  Please file the issue and assign it to me.  I'll 
> have a look at what needs to be added to support this.
>
> Mike
>
>
> ------------------------------------------------------------------------------
> Live Security Virtual Conference
> Exclusive live event will cover all the ways today's security and
> threat landscape has changed and how IT managers can respond. Discussions
> will include endpoint security, mobile security and the latest in malware
> threats. http://www.accelacomm.com/jaw/sfrnl04242012/114/50122263/
>
>
> _______________________________________________
> Matplotlib-users mailing list
> Mat...@li...
> https://lists.sourceforge.net/lists/listinfo/matplotlib-users

On 05/03/2012 09:48 AM, Benjamin Root wrote:
>
>
> On Wed, May 2, 2012 at 3:59 PM, David Erickson <der...@st... 
> <mailto:der...@st...>> wrote:
>
>     On 5/1/2012 5:41 AM, Benjamin Root wrote:
>>
>>
>>     On Tue, May 1, 2012 at 6:42 AM, David Erickson
>>     <der...@st... <mailto:der...@st...>> wrote:
>>
>>         Hi I need to use the symlog yscale in my graph, I have a lot
>>         of data
>>         that needs to be displayed linearly, with a small fraction at
>>         the upper
>>         end of the range that needs to be displayed in log scale due
>>         to its
>>         distance from the main data.  The symlog scale works great,
>>         however I'd
>>         like to adjust the actual vertical graph position where it
>>         crosses from
>>         linear to log (not the threshold), because right now only
>>         ~25% of the y
>>         space is being given to linear, and I'd like it to be more
>>         like 80%.  Is
>>         this possible?
>>
>>         Thanks!
>>         David
>>
>>
>>     Isn't linthreshy what you are looking for?  It denotes the range
>>     where the scale is linear.  So, if it is 25, then from -25 to 25
>>     the scale will be linear.  After 25, it will be log.
>>
>>     Maybe I am missing something in your description?
>>
>>     Cheers!
>>     Ben Root
>>
>
>     Hi Ben,
>     No unfortunately linthreshy only controls the crossover point from
>     linear to log scale, it does not give you control over where this
>     occurs on the figure's y coordinates.  I've attached a picture to
>     explain, currently the linear part of the graph is only being
>     given around 25% of the vertical space, I'd like to reverse that
>     and give linear ~75% and log at the top only the remaining 25%. 
>     Is this possible?  I've been digging around in the
>     SymmetricalLogScale and SymmetricalLogLocator classes and can't
>     even tell how this range is allocated.
>
>     Thanks in advance!!
>     -David
>
>
> David,
>
> Thanks, that is much clearer what you are looking for.  You are right, 
> I can't seem to find any sort of obvious way to get what you want.  
> Reading the docs for the SymLog scale indicates to me that the author 
> intended for the logrithmic portion to be most interesting and the 
> linear portion was only supposed to be a work-around the whole log(0) 
> issue.  Could you file an issue on the github page so that we can mark 
> it as a feature request?

Just getting to this after some unexpected absences.  Yes -- that is how 
symlog was originally intended, so there isn't currently any way to 
configure it.  Please file the issue and assign it to me.  I'll have a 
look at what needs to be added to support this.

Mike

On 05/03/2012 07:47 AM, Sandro Tosi wrote:
> Hi all,
>
> On Thu, May 3, 2012 at 1:25 PM, Michael Droettboom<md...@st...>  wrote:
>> Yeah, it definitely seems like there's a bit too much here.
> well, running "apt-get build-dep matplolib" you'll get all the
> dependecies needed to build all the part of matplotlib, both the
> module and the documentation; so it's surely a super-set of the
> packages needed only for doc... but... in order to build the doc, you
> need to build mpl module first, since it's used by the doc itself.

Sure.  I understand these are needed for both matplotlib itself and the 
docs.  I still think there's some extras, though.

>
>> I think some of
>> these dependencies are obsolete, in that we used to build extensions against
>> gtk and wx in order to transfer image data to those backends, but those are
>> no longer needed with current versions of those libraries.
>>
>> I looked up the direct dependencies of the package (since I think many on
>> the list you provided are probably secondary dependencies.)  I have some
>> comments below.  Maybe Sandro Tosi, who I know has done some Debian
>> packaging of matplotlib in the past,
> and I'm keep doing it these days :)

Great -- I didn't mean to imply you weren't still working on it ;)  
Thanks, as always, for all the work you've done on that.

>
>> libfreetype6-dev - legit
>> libgtk2.0-dev - **no longer necessary**
> mh, the GTK+ backend is built as an extensions, so it should require
> the development files to be built - f.e. I can see it "#include
> <pygtk/pygtk.h>"

Sorry -- I had misremembered.  It's Qt and Wx that used to have C/C++ 
extensions that are no longer required.  It seems we still need this for 
Gtk, so you're right -- this is a legitimate build dependency.

>
>> libpng-dev - legit
>> python-all-dbg (>= 2.4.4-6) - **not strictly necessary, but helpful for
>> debugging**
> that's because we're providing a debug-flavor extension, so we need
> the debug-enabled interpreters

Good.

>
>> python-all-dev (>= 2.3.5-7) - legit
>> python-configobj - **necessary only for a long abandoned experimental
>> version of matplotlib**
> I'll look if it can be removed - at a first glance it should be safe to rm it
>
>> python-dateutil - legit
>> python-epydoc - **obsolete**
> I'll see if it can be removed
>
>> python-gtk2-dev - **obsolete**
> see above

Agreed.

>
>> python-imaging - legit
>> python-numpy (>= 1:1.3.0) - legit
>> python-numpy-dbg (>= 1:1.3.0) - legit
>> python-pkg-resources - legit
>> python-qt4 - **not needed for build**
>> python-qt4-dev - **not needed for build**
> same as GTK+ backend

Nope.  There is no Qt C++ extension anymore (hasn't been for years).  
The only dependency is on python-qt4, and only at runtime, not build 
time, technically.

>
>> python-qt-dev - **obsolete**
> yeah it might now just be a transitional package towards
> python-qt4-dev - I'll check this
>
>> python-setuptools - legit
>> python-sphinx (>= 0.5.1) - to build the documentation
>> python-support (>= 1.0.0) - not sure what this is
> it's a debian tool to setup python modules/extensions for Debian systems

Ok.

>
>> python-tk (>= 2.5.2-1.1) - legit
>> python-traits (>= 2.0) - **not needed -- matplotlib doesn't use traits**
> yep, I'll remove it
>
>> python-tz - legit
>> python-wxgtk2.8 - **not needed for build**
>> python-wxgtk2.8-dbg - **not needed for build**
> I'll look into this

Same issue as for Qt/Qt4 -- there's no C code that links to it.

>
>> tcl8.5-dev - legit
>> texlive-fonts-recommended - needed for PDF doc build
>> texlive-latex-extra - needed for PDF doc build
>> texlive-latex-recommended - needed for PDF doc build
> that's probably where the big part of the download will come

Agreed.  But why would anyone not want TeX installed?  :)
>
>> tk8.5-dev - legit
>> zlib1g-dev - legit
> Thanks a lot for the review: it will help improving the package!
>
Thanks.

Mike

On Wed, May 2, 2012 at 3:59 PM, David Erickson <der...@st...>wrote:

>  On 5/1/2012 5:41 AM, Benjamin Root wrote:
>
>
>
> On Tue, May 1, 2012 at 6:42 AM, David Erickson <der...@st...>wrote:
>
>> Hi I need to use the symlog yscale in my graph, I have a lot of data
>> that needs to be displayed linearly, with a small fraction at the upper
>> end of the range that needs to be displayed in log scale due to its
>> distance from the main data.  The symlog scale works great, however I'd
>> like to adjust the actual vertical graph position where it crosses from
>> linear to log (not the threshold), because right now only ~25% of the y
>> space is being given to linear, and I'd like it to be more like 80%.  Is
>> this possible?
>>
>> Thanks!
>> David
>>
>>
> Isn't linthreshy what you are looking for?  It denotes the range where the
> scale is linear.  So, if it is 25, then from -25 to 25 the scale will be
> linear.  After 25, it will be log.
>
> Maybe I am missing something in your description?
>
> Cheers!
> Ben Root
>
>
> Hi Ben,
> No unfortunately linthreshy only controls the crossover point from linear
> to log scale, it does not give you control over where this occurs on the
> figure's y coordinates.  I've attached a picture to explain, currently the
> linear part of the graph is only being given around 25% of the vertical
> space, I'd like to reverse that and give linear ~75% and log at the top
> only the remaining 25%.  Is this possible?  I've been digging around in the
> SymmetricalLogScale and SymmetricalLogLocator classes and can't even tell
> how this range is allocated.
>
> Thanks in advance!!
> -David
>

David,

Thanks, that is much clearer what you are looking for.  You are right, I
can't seem to find any sort of obvious way to get what you want.  Reading
the docs for the SymLog scale indicates to me that the author intended for
the logrithmic portion to be most interesting and the linear portion was
only supposed to be a work-around the whole log(0) issue.  Could you file
an issue on the github page so that we can mark it as a feature request?

Thanks,
Ben Root

Yeah, it definitely seems like there's a bit too much here.  I think 
some of these dependencies are obsolete, in that we used to build 
extensions against gtk and wx in order to transfer image data to those 
backends, but those are no longer needed with current versions of those 
libraries.

I looked up the direct dependencies of the package (since I think many 
on the list you provided are probably secondary dependencies.)  I have 
some comments below.  Maybe Sandro Tosi, who I know has done some Debian 
packaging of matplotlib in the past, has some thoughts.

debhelper (>= 7) - probably needed to build a debian package
dvipng - to build the documentation
graphviz - to build the documentation
ipython - **seems unnecessary**
libfreetype6-dev - legit
libgtk2.0-dev - **no longer necessary**
libpng-dev - legit
python-all-dbg (>= 2.4.4-6) - **not strictly necessary, but helpful for debugging**
python-all-dev (>= 2.3.5-7) - legit
python-configobj - **necessary only for a long abandoned experimental version of matplotlib**
python-dateutil - legit
python-epydoc - **obsolete**
python-gtk2-dev - **obsolete**
python-imaging - legit
python-numpy (>= 1:1.3.0) - legit
python-numpy-dbg (>= 1:1.3.0) - legit
python-pkg-resources - legit
python-qt4 - **not needed for build**
python-qt4-dev - **not needed for build**
python-qt-dev - **obsolete**
python-setuptools - legit
python-sphinx (>= 0.5.1) - to build the documentation
python-support (>= 1.0.0) - not sure what this is
python-tk (>= 2.5.2-1.1) - legit
python-traits (>= 2.0) - **not needed -- matplotlib doesn't use traits**
python-tz - legit
python-wxgtk2.8 - **not needed for build**
python-wxgtk2.8-dbg - **not needed for build**
tcl8.5-dev - legit
texlive-fonts-recommended - needed for PDF doc build
texlive-latex-extra - needed for PDF doc build
texlive-latex-recommended - needed for PDF doc build
tk8.5-dev - legit
zlib1g-dev - legit


Mike


On 05/02/2012 07:09 PM, Benjamin Root wrote:
>
>
> On Wednesday, May 2, 2012, Alexis Praga wrote:
>
>     Here is the result of "apt-get build-dep python-matplotlib" :
>
>     The following NEW packages will be installed:
>      blt debhelper doc-base docbook-xsl docbook-xsl-doc-html dvipng
>     global graphviz html2text intltool-debian ipython javascript-common
>     libatk1.0-dev libcairo2-dev libcgraph5 libfreezethaw-perl
>      libgtk2.0-dev libgvpr1 libjs-jquery libmail-sendmail-perl
>     libmldbm-perl libpango1.0-dev libphonon4 libpixman-1-dev
>     libqt4-assistant libqt4-help libqt4-scripttools libqt4-svg libqt4-test
>     libqt4-webkit
>      libqt4-xmlpatterns libsys-hostname-long-perl libwxbase2.8-dbg
>     libwxgtk2.8-dbg libxcb-render-util0-dev libxcb-render0-dev
>     libxcomposite-dev libxcursor-dev libxdamage-dev libxfixes-dev
>     libxi-dev
>      libxrandr-dev libxss-dev po-debconf preview-latex-style python-all
>     python-all-dbg python-all-dev python-apptools python-dateutil
>     python-dbg python-docutils python-enthoughtbase python-epydoc
>      python-foolscap python-gobject-dev python-gtk2-dev python-gtk2-doc
>     python-jinja2 python-lxml python-numpy-dbg python-pygments
>     python-qt-dev python-qt4 python-qt4-dev python-roman python-setuptools
>      python-sip python-sip-dev python-sphinx python-tk python-traits
>     python-twisted-names python-tz python-wxgtk2.8 python-wxgtk2.8-dbg
>     python-wxversion python2.5 python2.5-dbg python2.5-dev
>      python2.5-minimal python2.6-dbg tcl8.5-dev texlive-fonts-recommended
>     texlive-fonts-recommended-doc texlive-latex-extra
>     texlive-latex-extra-doc texlive-pictures texlive-pictures-doc texpower
>      texpower-manual tipa tk8.5-dev wwwconfig-common wx2.8-headers
>     x11proto-composite-dev x11proto-damage-dev x11proto-fixes-dev
>     x11proto-randr-dev x11proto-scrnsaver-dev
>     0 upgraded, 100 newly installed, 0 to remove and 2 not upgraded.
>     Need to get 298 MB/369 MB of archives.
>     After this operation, 818 MB of additional disk space will be used.
>
>     Now, with "apt-rdepends python-matplotlib | head -n 24", I can have
>     the "true" dependencies :
>     python-matplotlib
>      Depends: libatk1.0-0 (>= 1.29.3)
>      Depends: libc6 (>= 2.3.6-6~)
>      Depends: libcairo2 (>= 1.2.4)
>      Depends: libfontconfig1 (>= 2.8.0)
>      Depends: libfreetype6 (>= 2.2.1)
>      Depends: libgcc1 (>= 1:4.1.1)
>      Depends: libglib2.0-0 (>= 2.16.0)
>      Depends: libgtk2.0-0 (>= 2.8.0)
>      Depends: libpango1.0-0 (>= 1.14.0)
>      Depends: libpng12-0 (>= 1.2.13-4)
>      Depends: libstdc++6 (>= 4.4.0)
>      Depends: python (>= 2.5)
>      Depends: python-cairo
>      Depends: python-dateutil
>      Depends: python-gobject
>      Depends: python-matplotlib-data (>= 0.99.3-1)
>      Depends: python-numpy (>= 1:1.3.0)
>      Depends: python-pyparsing
>      Depends: python-support (>= 0.90.0)
>      Depends: python-tz
>      Depends: tcl8.5 (>= 8.5.0)
>      Depends: tk8.5 (>= 8.5.0)
>      Depends: zlib1g (>= 1:1.1.4)
>
>     --
>
>     Alexis Praga
>
>
> Interesting, ipython is not a dependency of mpl. This repo is 
> packaging mpl incorrectly.
>
> Ben Root
>
>
> ------------------------------------------------------------------------------
> Live Security Virtual Conference
> Exclusive live event will cover all the ways today's security and
> threat landscape has changed and how IT managers can respond. Discussions
> will include endpoint security, mobile security and the latest in malware
> threats. http://www.accelacomm.com/jaw/sfrnl04242012/114/50122263/
>
>
> _______________________________________________
> Matplotlib-users mailing list
> Mat...@li...
> https://lists.sourceforge.net/lists/listinfo/matplotlib-users