[Yt-dev] Call for testing: Freetype module build system

[John Wise]

Hi Matt,

I managed to fix this by installing my own copy of libpng.  Previously I was using the system's copy in /usr.  But there's also a copy of freetype2 in /usr.  So the include paths had two copies of freetype2 in them with the system-wide (faulty) one taking precedence.  I'm not sure why this didn't cause me problems with matplotlib.

John

On 14 Dec 2010, at 14:12, Matthew Turk wrote:

> Hi John,
> 
> Okay, I think this might be related to the way FreeType2 does its
> includes, which I think is a bit odd, and which I may not have
> implemented correctly in yt/utilities/setup.py.  Is it finding the
> correct .h files, and then it's not able to identify the symbols?  (At
> compilation, or during link?)  Can you fiddle with the actual freetype
> includes in that setup.py, and then see if there's some kind of
> logical sense I am missing with the build?
> 
> -Matt
> 
> On Tue, Dec 14, 2010 at 11:10 AM, John Wise <jwise at astro.princeton.edu> wrote:
>> It still gives the same error.
>> 
>> C compiler: icc -pthread -fno-strict-aliasing -g -O2 -DNDEBUG -g -O3 -Wall -Wstrict-prototypes -fPIC
>> 
>> compile options: '-DPNG_SETJMP_NOT_SUPPORTED=True -Iyt/utilities/_amr_utils/ -I/usr/include -I/u/jhwise/local/include -I/u/jhwise/local/include/freetype2 -I/u/jhwise/local/lib/python2.7/site-packages/numpy/core
>> /include -I/u/jhwise/local/include/python2.7 -c'
>> icc: yt/utilities/_amr_utils/FixedInterpolator.c
>> icc: command line warning #10120: overriding '-O2' with '-O3'
>> [...]
>> yt/utilities/amr_utils.c(1245): error: identifier "FT_Library" is undefined
>> 
>> There are similar errors when FT_Face, FT_GlyphSlot, etc. are undefined.
>> 
>> John
>> 
>> 
>> On 14 Dec 2010, at 14:02, Matthew Turk wrote:
>> 
>>> Hi all,
>>> 
>>> Thanks for the (so far!) positive (mostly!) feedback.
>>> 
>>> John -- you should be able to set the base directory in freetype.cfg,
>>> just like png.cfg and hdf5.cfg.  If you give that a shot, can you tell
>>> me if it works?
>>> 
>>> Also, good to know about the 0.13 thing; cython 0.13 is now part of
>>> the installation script, but with any luck we can just ship the .c
>>> files for the most part.
>>> 
>>> Sam -- I agree, we should have more unified image handling.  As it
>>> stands, the bitmapping and whatnot is a bit annoying.  I think in
>>> conjunction with the plot objects (and FRBs) we currently have, an
>>> ImageBuffer object would be nice.  It could be created from floating
>>> point values, and then with min/max/colormap properties it could have
>>> a .bitmap property that would return a bitmap.  Or something like
>>> that.  I'm still thinking on it, but I believe we could do something
>>> useful with images.
>>> 
>>> -Matt
>>> 
>>> On Tue, Dec 14, 2010 at 10:51 AM, John Wise <jwise at astro.princeton.edu> wrote:
>>>> On 14 Dec 2010, at 13:49, John Wise wrote:
>>>> 
>>>>> This is great.  It installed with one problem (not related to the freetype additions) on OSX 10.6 with yt installed from the install script.  I still had cython 0.12.1, and it gave me this error
>>>> 
>>>> I forgot to mention that this error went away when I installed cython 0.13.
>>>> 
>>>> John
>>>> _______________________________________________
>>>> Yt-dev mailing list
>>>> Yt-dev at lists.spacepope.org
>>>> http://lists.spacepope.org/listinfo.cgi/yt-dev-spacepope.org
>>>> 
>>> _______________________________________________
>>> Yt-dev mailing list
>>> Yt-dev at lists.spacepope.org
>>> http://lists.spacepope.org/listinfo.cgi/yt-dev-spacepope.org
>> 
>> _______________________________________________
>> Yt-dev mailing list
>> Yt-dev at lists.spacepope.org
>> http://lists.spacepope.org/listinfo.cgi/yt-dev-spacepope.org
>> 
> _______________________________________________
> Yt-dev mailing list
> Yt-dev at lists.spacepope.org
> http://lists.spacepope.org/listinfo.cgi/yt-dev-spacepope.org