pmxbot IRC Log Viewer

[15:28:55] <wiredfool> jleclanche: got your email ccd' from alex

[15:29:00] <jleclanche> wiredfool: hey!

[15:29:14] <wiredfool> also, I'm pretty sure that the gbr plugin hasn't worked since the PIL days

[15:29:22] <jleclanche> yeah that figures

[15:29:29] <jleclanche> I mean it's not a very useful plugin

[15:29:33] <wiredfool> not that it's stopped code health and flake8 fixes

[15:30:09] <wiredfool> yah, well, it's works _now_

[15:30:27] <wiredfool> and it's not a very hard plugin

[15:31:57] <wiredfool> useful excersize for figuring out wtfh on the plugins

[15:32:09] <jleclanche> mm

[15:32:27] <travis-ci> wiredfool/Pillow#746 (gbr - 25428af : wiredfool): The build failed.

[15:32:27] <travis-ci> Change view : https://github.com/wiredfool/Pillow/compare/ccd9d4b489e0^...25428afb7084

[15:32:27] <travis-ci> Build details : https://travis-ci.org/wiredfool/Pillow/builds/101083484

[15:33:35] <jleclanche> wiredfool: do you think the dds PR can be merged now?

[15:33:40] <jleclanche> it just needs some squashing

[15:33:51] <wiredfool> let me look

[15:37:27] <wiredfool> My one reservation at this point is that the tests don't test for correctness. They test that a file is returned with the right metadata and the right plugin type.

[15:37:46] <wiredfool> Plugin Type is not something I care about, as long as it's an image.

[15:38:20] <wiredfool> But if we replace this with a C implementation, then there's no test that we've done it right. Nor if we run this on a Big endian machine.

[15:38:41] <wiredfool> (but coverage is an excellend 9something percent)

[15:39:01] <wiredfool> This is how shit breaks.

[15:41:25] <jleclanche> wiredfool: part of why that is is that we barely cover dds

[15:41:35] <jleclanche> dxt1 and dxt5 are the major dds users

[15:41:50] <jleclanche> but going by implementation, that's like 3% of dds

[15:43:16] <wiredfool> so, are you saying that it's a fragmented format?

[15:43:24] <jleclanche> dds is a container format

[15:43:25] <wiredfool> and we're covering 3% of the usecases?

[15:43:42] <jleclanche> it describes the pixel format used essentially

[15:43:55] <wiredfool> and we is...

[15:43:57] <jleclanche> most of dds is not actually necessary, it's just optimizations

[15:44:06] <jleclanche> we = the dds plugin

[15:46:26] <travis-ci> wiredfool/Pillow#747 (gbr - 00dc657 : wiredfool): The build has errored.

[15:46:26] <travis-ci> Change view : https://github.com/wiredfool/Pillow/compare/25428afb7084...00dc65782488

[15:46:26] <travis-ci> Build details : https://travis-ci.org/wiredfool/Pillow/builds/101084915

[15:47:46] <wiredfool> ok

[15:53:42] <wiredfool> jleclanche: not getting good results from the correctness test

[15:54:07] <jleclanche> i dont know about the test :) it might be incorrect

[15:54:24] <wiredfool> it's the one that I'm running

[15:54:56] <wiredfool> File looks garbled when running im.show()

[15:55:10] <jleclanche> what's the input file format?

[15:55:11] <wiredfool> where im=Image.open('Tests/images/dxt1-rgb-4bbp-noalpha_MipMaps-1.dds')

[15:55:18] <jleclanche> hmm

[15:55:34] <jleclanche> yeah I don't know about it, but I use the plugin I did in my dds files

[15:55:55] <jleclanche> wiredfool: give the dds.zip I attached to the issue tracker a shot

[15:56:05] <wiredfool> ok

[15:59:27] <wiredfool> ok, in imagemagick's display, I see a world.

[15:59:36] <wiredfool> from pillow, I see noise

[16:00:27] <wiredfool> similar results to the ones in the PR

[16:05:37] <jleclanche> huh

[16:05:40] <jleclanche> I use feh

[16:06:00] <jleclanche> i think that uses the same lib as imagemagick

[16:06:13] <jleclanche> actually it doesnt, libimaging

[16:09:08] <wiredfool> https://github.com/python-pillow/Pillow/pull/1654 , see the last comment for what I'm seeing.

[16:15:45] <travis-ci> wiredfool/Pillow#748 (pr_1650 - 13f2d22 : wiredfool): The build failed.

[16:15:45] <travis-ci> Change view : https://github.com/wiredfool/Pillow/compare/8ece7f6c9e5b^...13f2d227009a

[16:15:45] <travis-ci> Build details : https://travis-ci.org/wiredfool/Pillow/builds/101093554

[16:28:05] <jleclanche> wiredfool: will you look into this error?

[16:28:12] <jleclanche> I'm not seeing it so I'm not sure if I can help

[16:28:36] <jleclanche> a garbled output like this makes very little sense

[16:29:43] <jleclanche> it's like you're reading random noncolor data

[16:29:51] <wiredfool> are you developing on a bigendian machine?

[16:30:08] <jleclanche> hmm

[16:30:22] <jleclanche> good question. I'm on a LE machine but it's through the unity3d layer

[16:30:27] <jleclanche> which is big endian I *believe*

[16:30:29] <jleclanche> let me check

[16:31:01] <jleclanche> nope, it's little endian - the header only is big endian

[16:32:41] <wiredfool> if you pull my pr_1650 branch, do you get the error that the content is different?

[16:34:05] <jleclanche> wiredfool: can you link your branch? github is being uncooperative

[16:34:13] <jleclanche> your remote i mean

[16:35:09] <wiredfool> https://github.com/wiredfool/Pillow.git

[16:40:02] <jleclanche> wiredfool: what do I run?

[16:40:16] <jleclanche> test_file_dds.py:3: in <module>

[16:40:16] <jleclanche> from PIL import Image, DdsImagePlugin

[16:40:17] <jleclanche> E ImportError: cannot import name 'DdsImagePlugin'

[16:40:18] <jleclanche> bleh

[16:40:55] <wiredfool> I use a virtualenv, then make install && make test

[16:41:05] <wiredfool> or ./test-installed.py -s Tests/test_file_dds.py

[16:41:16] <wiredfool> for just the one test

[16:42:08] <wiredfool> travis is failing at the same test, so it's not _just_ me

[16:43:50] <jleclanche> I have one error

[16:43:56] <jleclanche> FAIL: Check DXT5 images can be opened

[16:43:56] <jleclanche> File "/home/adys/tmp/Pillow/Tests/helper.py", line 76, in assert_image_equal

[16:43:56] <jleclanche> self.fail(msg or "got different content")

[16:43:57] <jleclanche> AssertionError: got different content

[16:44:57] <wiredfool> so, if you open that image with Pillow and use im.show(), does it look ok?

[16:45:32] <jleclanche> dxt5-argb-8bbp-interpolatedalpha_MipMaps-1.dds ?

[16:45:38] <wiredfool> (or save as png and then view in some other editor)

[16:45:39] <wiredfool> yes

[16:45:55] <jleclanche> yeah it looks fine

[16:47:04] <jleclanche> I unpack everything in LE mode

[16:47:10] <jleclanche> so it cant be failing on endians I dont think

[16:47:54] <jleclanche> but it looks like the original yes

[16:48:23] <wiredfool> from what I see of this format, it should be exactly the same

[16:48:29] <wiredfool> i.e., not lossy

[16:49:23] <jleclanche> wiredfool: there's jpg format but that's super rare and not implemented in the version i gave you

[16:49:35] <jleclanche> everything else is nonlossy

[17:03:14] <wiredfool> jleclanche: it doesn't look wrong from examination. I'd have to trace through it to see if it's doing what I think it is through

[17:03:22] <wiredfool> but now, I've got to make dinner

[17:04:15] <jleclanche> yeah Im off for a while as well

[17:04:18] <jleclanche> ill be back in a few

[18:13:00] <travis-ci> wiredfool/Pillow#749 (sphinx-travis - 8d518a3 : wiredfool): The build was canceled.

[18:13:00] <travis-ci> Change view : https://github.com/wiredfool/Pillow/compare/sphinx-travis

[18:13:00] <travis-ci> Build details : https://travis-ci.org/wiredfool/Pillow/builds/101121960

[18:33:52] <travis-ci> wiredfool/Pillow#750 (sphinx-travis - f09508f : wiredfool): The build has errored.

[18:33:52] <travis-ci> Change view : https://github.com/wiredfool/Pillow/compare/8d518a3f4d7d...f09508f17f30

[18:33:52] <travis-ci> Build details : https://travis-ci.org/wiredfool/Pillow/builds/101122862

[18:36:13] <travis-ci> wiredfool/Pillow#751 (gbr - d047349 : wiredfool): The build passed.

[18:36:13] <travis-ci> Change view : https://github.com/wiredfool/Pillow/compare/00dc65782488...d04734995816

[18:36:13] <travis-ci> Build details : https://travis-ci.org/wiredfool/Pillow/builds/101124186

[18:36:51] <travis-ci> wiredfool/Pillow#751 (gbr - d047349 : wiredfool): The build passed.

[18:36:51] <travis-ci> Change view : https://github.com/wiredfool/Pillow/compare/00dc65782488...d04734995816

[18:36:51] <travis-ci> Build details : https://travis-ci.org/wiredfool/Pillow/builds/101124186

[19:15:57] <travis-ci> wiredfool/Pillow#750 (sphinx-travis - f09508f : wiredfool): The build passed.

[19:15:57] <travis-ci> Change view : https://github.com/wiredfool/Pillow/compare/8d518a3f4d7d...f09508f17f30

[19:15:57] <travis-ci> Build details : https://travis-ci.org/wiredfool/Pillow/builds/101122862

[21:11:48] <travis-ci> python-pillow/Pillow#2808 (master - 66c99b9 : wiredfool): The build has errored.

[21:11:48] <travis-ci> Change view : https://github.com/python-pillow/Pillow/compare/f57e295ce5ff...66c99b9d202c

[21:11:48] <travis-ci> Build details : https://travis-ci.org/python-pillow/Pillow/builds/101155173

[21:33:31] <travis-ci> radarhere/Pillow#834 (master - 66c99b9 : wiredfool): The build passed.

[21:33:31] <travis-ci> Change view : https://github.com/radarhere/Pillow/compare/f57e295ce5ff...66c99b9d202c

[21:33:31] <travis-ci> Build details : https://travis-ci.org/radarhere/Pillow/builds/101161001

[21:35:04] <travis-ci> radarhere/Pillow#834 (master - 66c99b9 : wiredfool): The build passed.

[21:35:04] <travis-ci> Change view : https://github.com/radarhere/Pillow/compare/f57e295ce5ff...66c99b9d202c

[21:35:04] <travis-ci> Build details : https://travis-ci.org/radarhere/Pillow/builds/101161001

[21:52:40] <wiredfool> jleclanche: got it

[21:53:37] <wiredfool> two issues: 1, bytes() == str() in py2.x, so that bytes([r,g,b,alpha]) turned into the string '[0, 0, 0, 0]', of which you were taking the first 4 characters.

[21:55:30] <wiredfool> The second is that imagemagick has slightly different bits on the dxt5 image, it's interpreting the color as (00, 6d, ff, ff), where the color should be (00,6c,f8,ff), at least according to the non-imagemagick'd target image for the dxt1 image.

[22:07:35] <travis-ci> python-pillow/Pillow#2808 (master - 66c99b9 : wiredfool): The build passed.

[22:07:35] <travis-ci> Change view : https://github.com/python-pillow/Pillow/compare/f57e295ce5ff...66c99b9d202c

[22:07:35] <travis-ci> Build details : https://travis-ci.org/python-pillow/Pillow/builds/101155173

[22:37:42] <travis-ci> wiredfool/Pillow#753 (pr_1650 - 18d48dc : wiredfool): The build has errored.

[22:37:42] <travis-ci> Change view : https://github.com/wiredfool/Pillow/compare/13f2d227009a...18d48dc66521

[22:37:42] <travis-ci> Build details : https://travis-ci.org/wiredfool/Pillow/builds/101169528

[23:59:10] <travis-ci> radarhere/Pillow#835 (redirect - 8a27dd3 : Andrew Murray): The build passed.

[23:59:10] <travis-ci> Change view : https://github.com/radarhere/Pillow/commit/8a27dd3b0877

[23:59:10] <travis-ci> Build details : https://travis-ci.org/radarhere/Pillow/builds/101189325

Log file Viewer

Help | Karma | Search:

#pil logs for Friday the 8th of January, 2016