Log file Viewer

[04:56:27] <wiredfool> iElectric: sorry, mostly idle on weekends. I'm normally around business hours, pacific time zone.

[07:52:42] <iElectric> wiredfool: sure

[07:53:32] <iElectric> wiredfool: I was thinking how to solve the problem that Pillow doesn't obey

[07:53:37] <iElectric> CFLAGS="-I..."

[07:53:43] <iElectric> and LDFLAGS="-L..."

[07:54:08] <iElectric> since currently distutils machinery just passes that to the compiler

[07:54:27] <iElectric> but pillow has lots of logic to inspect if libraries are installed on the system

[07:54:56] <iElectric> so I'm worndering why not just try to compile those libs and silently fail in case of error?

[14:01:45] <ttback> hello

[14:01:47] <ttback> just a question

[14:02:00] <ttback> is there any built-in multiprocessing support inside pillow?

[14:02:12] <ttback> or it focuses on just getting the process part correctly

[16:01:49] <wiredfool> iElectric: We've been generally moving towards explicit failures. I anticipate that we're going to mane at least a few of the 'optional' libraries required soon, since frex jpeg and png functionality are sort of required to be considered a minimally competent image library these days.

[16:02:48] <wiredfool> I think that we've had patches looking at the cflags/ldflags, I know we've had a lot of patches that add to the long list of where we're looking for libraries.

[16:37:07] <iElectric> wiredfool: it doesn't work currently

[16:40:31] <iElectric> wiredfool: I'll supply a patch then

[16:40:55] <wiredfool> looks like we haven't doine that.

[16:41:05] <wiredfool> the only cflags ref is adding -wall

[16:49:17] <iElectric> wiredfool: why wouldn't you try to compile the extensions?

[16:49:29] <iElectric> guessing support is really hard and more or less impossible

[16:49:35] <iElectric> knob on a knob

[16:50:52] <iElectric> sqlalchemy and jinja2 do that

[16:50:57] <iElectric> (just try to compile)

[16:54:31] <wiredfool> take a look at encode.c, decode.c, and _imaging.c

[16:54:53] <wiredfool> especially, HAVE_LIBZ, HAVE_LIBJPEG, etc

[16:55:25] <wiredfool> Basically, either we rearchitect how the c extension is put together

[16:55:53] <wiredfool> or we do something like ./configure where we test compile

[16:56:11] <wiredfool> or we have to figure out the exact combination of what's there by trial and error

[16:57:44] <wiredfool> at least lcms and freetype are relatively easy to split out.

[16:58:28] <wiredfool> the newer additions are a little more modular than the older ones.

[18:48:04] <iElectric> hmmmm

[19:12:52] <iElectric> wiredfool: ok, I guess I'm not ready to rewrite that part

[19:12:55] <iElectric> so I'll shut up

[19:13:11] <iElectric> I'll try to provide a patch that takes in account CFLAGS/LDFLAGS

[19:13:40] <wiredfool> ok. the setup stuff has been a pain for a lot of people, and some of it has been around far longer than one would really believe possible

[19:15:50] <iElectric> although the patch will be ugly since there is no easy way to fix that this way

[19:45:43] <iElectric> there

[19:45:43] <iElectric> https://github.com/python-imaging/Pillow/pull/368

[19:45:45] <iElectric> wiredfool:

[19:57:43] <wiredfool> I see it.

[19:59:22] <wiredfool> hey aclark

[20:00:27] <wiredfool> iElectric: I want to test your patch prior to merging

[20:02:39] <aclark> wiredfool: cool i'll wait

[20:02:46] <aclark> wiredfool: or you can just merge when you are done

[20:03:07] <wiredfool> I also want to pull out any "string string".split refs and use lists or tuples

[20:03:12] <wiredfool> in setup.py

[20:04:23] <wiredfool> and.. looking at setup.py, there's a reference to _imagingtiff.c . no clue where that came from, unless Fredrik was working on that for PIL 1.1.8

[20:04:36] <wiredfool> It's pre-github at least

[20:06:18] <iElectric> wiredfool: sure

[20:13:23] <aclark> so many bodies, only Fredrik knows the burial location ;-)