Discussion:
02/02: gnu: python-pillow: Update to 4.3.0.
(too old to reply)
Mark H Weaver
2017-12-24 17:57:16 UTC
Permalink
kkebreau pushed a commit to branch master
in repository guix.
commit 9e3a8ed0ebc5b4095f1b64d85fd56ce7fb636580
Date: Mon Dec 4 17:56:37 2017 -0500
gnu: python-pillow: Update to 4.3.0.
This new version seems to deterministically fail its test suite on i686:

https://hydra.gnu.org/build/2400760
https://hydra.gnu.org/build/2400819

I restarted both of these, and they each failed twice. The same builds
(python-pillow and python2-pillow) also failed on armhf, although as I
write this they have not yet completed their second build attempts.

Altogether, this caused around 150 new dependency failures:

https://hydra.gnu.org/eval/109866#tabs-now-fail

See below for the tail of one of the i686 build logs.

Could you take a look?

Mark


--8<---------------cut here---------------start------------->8---
FAIL: TestFontPcf.test_high_characters
----------------------------------------------------------------------
Traceback (most recent call last):
File "/tmp/guix-build-python-pillow-4.3.0.drv-0/Pillow-4.3.0/Tests/test_font_pcf.py", line 62, in test_high_characters
self._test_high_characters(message.encode('latin1'))
File "/tmp/guix-build-python-pillow-4.3.0.drv-0/Pillow-4.3.0/Tests/test_font_pcf.py", line 55, in _test_high_characters
self.assert_image_equal(image, compare)
File "/tmp/guix-build-python-pillow-4.3.0.drv-0/Pillow-4.3.0/Tests/helper.py", line 85, in assert_image_equal
msg or "got size %r, expected %r" % (a.size, b.size))
AssertionError: Tuples differ: (775, 22) != (765, 22)

First differing element 0:
775
765

- (775, 22)
? ^

+ (765, 22)
? ^
: got size (775, 22), expected (765, 22)
-------------------- >> begin captured logging << --------------------
PIL.PngImagePlugin: DEBUG: STREAM b'IHDR' 16 13
PIL.PngImagePlugin: DEBUG: STREAM b'IDAT' 41 2644
PIL.PngImagePlugin: DEBUG: STREAM b'IHDR' 16 13
PIL.PngImagePlugin: DEBUG: STREAM b'IDAT' 41 1400
PIL.PngImagePlugin: DEBUG: STREAM b'IHDR' 16 13
PIL.PngImagePlugin: DEBUG: STREAM b'IDAT' 41 2644
PIL.PngImagePlugin: DEBUG: STREAM b'IHDR' 16 13
PIL.PngImagePlugin: DEBUG: STREAM b'IDAT' 41 1400
--------------------- >> end captured logging << ---------------------

----------------------------------------------------------------------
Ran 1085 tests in 4.238s

FAILED (SKIP=116, failures=1)
phase `check-installed' failed after 8.9 seconds
builder for `/gnu/store/74g32xnyc7zzang94ji4jh1xbiqmcjks-python-pillow-4.3.0.drv' failed with exit code 1
@ build-failed /gnu/store/74g32xnyc7zzang94ji4jh1xbiqmcjks-python-pillow-4.3.0.drv - 1 builder for `/gnu/store/74g32xnyc7zzang94ji4jh1xbiqmcjks-python-pillow-4.3.0.drv' failed with exit code 1
--8<---------------cut here---------------end--------------->8---
Mark H Weaver
2017-12-24 17:59:12 UTC
Permalink
Hi Kei,
kkebreau pushed a commit to branch master
in repository guix.
commit 9e3a8ed0ebc5b4095f1b64d85fd56ce7fb636580
Date: Mon Dec 4 17:56:37 2017 -0500
gnu: python-pillow: Update to 4.3.0.
This new version seems to deterministically fail its test suite on i686:

https://hydra.gnu.org/build/2400760
https://hydra.gnu.org/build/2400819

I restarted both of these, and they each failed again the second time.
The same builds (python-pillow and python2-pillow) also failed on armhf,
although as I write this they have not yet completed their second build
attempts.

Altogether, this caused around 150 new dependency failures:

https://hydra.gnu.org/eval/109866#tabs-now-fail

See below for the tail of one of the i686 build logs.

Could you take a look?

Mark


--8<---------------cut here---------------start------------->8---
FAIL: TestFontPcf.test_high_characters
----------------------------------------------------------------------
Traceback (most recent call last):
File "/tmp/guix-build-python-pillow-4.3.0.drv-0/Pillow-4.3.0/Tests/test_font_pcf.py", line 62, in test_high_characters
self._test_high_characters(message.encode('latin1'))
File "/tmp/guix-build-python-pillow-4.3.0.drv-0/Pillow-4.3.0/Tests/test_font_pcf.py", line 55, in _test_high_characters
self.assert_image_equal(image, compare)
File "/tmp/guix-build-python-pillow-4.3.0.drv-0/Pillow-4.3.0/Tests/helper.py", line 85, in assert_image_equal
msg or "got size %r, expected %r" % (a.size, b.size))
AssertionError: Tuples differ: (775, 22) != (765, 22)

First differing element 0:
775
765

- (775, 22)
? ^

+ (765, 22)
? ^
: got size (775, 22), expected (765, 22)
-------------------- >> begin captured logging << --------------------
PIL.PngImagePlugin: DEBUG: STREAM b'IHDR' 16 13
PIL.PngImagePlugin: DEBUG: STREAM b'IDAT' 41 2644
PIL.PngImagePlugin: DEBUG: STREAM b'IHDR' 16 13
PIL.PngImagePlugin: DEBUG: STREAM b'IDAT' 41 1400
PIL.PngImagePlugin: DEBUG: STREAM b'IHDR' 16 13
PIL.PngImagePlugin: DEBUG: STREAM b'IDAT' 41 2644
PIL.PngImagePlugin: DEBUG: STREAM b'IHDR' 16 13
PIL.PngImagePlugin: DEBUG: STREAM b'IDAT' 41 1400
--------------------- >> end captured logging << ---------------------

----------------------------------------------------------------------
Ran 1085 tests in 4.238s

FAILED (SKIP=116, failures=1)
phase `check-installed' failed after 8.9 seconds
builder for `/gnu/store/74g32xnyc7zzang94ji4jh1xbiqmcjks-python-pillow-4.3.0.drv' failed with exit code 1
@ build-failed /gnu/store/74g32xnyc7zzang94ji4jh1xbiqmcjks-python-pillow-4.3.0.drv - 1 builder for `/gnu/store/74g32xnyc7zzang94ji4jh1xbiqmcjks-python-pillow-4.3.0.drv' failed with exit code 1
--8<---------------cut here---------------end--------------->8---
Kei Kebreau
2017-12-24 19:08:51 UTC
Permalink
Post by Mark H Weaver
Hi Kei,
kkebreau pushed a commit to branch master
in repository guix.
commit 9e3a8ed0ebc5b4095f1b64d85fd56ce7fb636580
Date: Mon Dec 4 17:56:37 2017 -0500
gnu: python-pillow: Update to 4.3.0.
https://hydra.gnu.org/build/2400760
https://hydra.gnu.org/build/2400819
I restarted both of these, and they each failed again the second time.
The same builds (python-pillow and python2-pillow) also failed on armhf,
although as I write this they have not yet completed their second build
attempts.
https://hydra.gnu.org/eval/109866#tabs-now-fail
See below for the tail of one of the i686 build logs.
Could you take a look?
Mark
FAIL: TestFontPcf.test_high_characters
----------------------------------------------------------------------
File "/tmp/guix-build-python-pillow-4.3.0.drv-0/Pillow-4.3.0/Tests/test_font_pcf.py", line 62, in test_high_characters
self._test_high_characters(message.encode('latin1'))
File "/tmp/guix-build-python-pillow-4.3.0.drv-0/Pillow-4.3.0/Tests/test_font_pcf.py", line 55, in _test_high_characters
self.assert_image_equal(image, compare)
File "/tmp/guix-build-python-pillow-4.3.0.drv-0/Pillow-4.3.0/Tests/helper.py", line 85, in assert_image_equal
msg or "got size %r, expected %r" % (a.size, b.size))
AssertionError: Tuples differ: (775, 22) != (765, 22)
775
765
- (775, 22)
? ^
+ (765, 22)
? ^
: got size (775, 22), expected (765, 22)
-------------------- >> begin captured logging << --------------------
PIL.PngImagePlugin: DEBUG: STREAM b'IHDR' 16 13
PIL.PngImagePlugin: DEBUG: STREAM b'IDAT' 41 2644
PIL.PngImagePlugin: DEBUG: STREAM b'IHDR' 16 13
PIL.PngImagePlugin: DEBUG: STREAM b'IDAT' 41 1400
PIL.PngImagePlugin: DEBUG: STREAM b'IHDR' 16 13
PIL.PngImagePlugin: DEBUG: STREAM b'IDAT' 41 2644
PIL.PngImagePlugin: DEBUG: STREAM b'IHDR' 16 13
PIL.PngImagePlugin: DEBUG: STREAM b'IDAT' 41 1400
--------------------- >> end captured logging << ---------------------
----------------------------------------------------------------------
Ran 1085 tests in 4.238s
FAILED (SKIP=116, failures=1)
phase `check-installed' failed after 8.9 seconds
builder for `/gnu/store/74g32xnyc7zzang94ji4jh1xbiqmcjks-python-pillow-4.3.0.drv' failed with exit code 1
@ build-failed /gnu/store/74g32xnyc7zzang94ji4jh1xbiqmcjks-python-pillow-4.3.0.drv - 1 builder for `/gnu/store/74g32xnyc7zzang94ji4jh1xbiqmcjks-python-pillow-4.3.0.drv' failed with exit code 1
I'm investigating now. Thanks for the heads up.
Mark H Weaver
2017-12-24 21:57:17 UTC
Permalink
Post by Kei Kebreau
Post by Mark H Weaver
kkebreau pushed a commit to branch master
in repository guix.
commit 9e3a8ed0ebc5b4095f1b64d85fd56ce7fb636580
Date: Mon Dec 4 17:56:37 2017 -0500
gnu: python-pillow: Update to 4.3.0.
https://hydra.gnu.org/build/2400760
https://hydra.gnu.org/build/2400819
I restarted both of these, and they each failed again the second time.
The same builds (python-pillow and python2-pillow) also failed on armhf,
although as I write this they have not yet completed their second build
attempts.
The (new) failure on armhf also appears to be consistent, although
different from the i686 failure. Instead of a failed test, the build
times out after 1 hour of silence during the test suite:

https://hydra.gnu.org/build/2400971

--8<---------------cut here---------------start------------->8---
Running selftest:
--- 57 tests passed.
...SSSSS.......S..SS.....................................................SS...................................SSSSS..........................................................................................................................S....S.S.................S....................S.............................................................................S.building of `/gnu/store/cjinxvws27bwdwn7n2fab3k10had6y2p-python2-pillow-4.3.0.drv' timed out after 3600 seconds of silence
@ build-failed /gnu/store/cjinxvws27bwdwn7n2fab3k10had6y2p-python2-pillow-4.3.0.drv - timeout
--8<---------------cut here---------------end--------------->8---
Post by Kei Kebreau
I'm investigating now. Thanks for the heads up.
Thanks!
Mark
Kei Kebreau
2017-12-24 23:16:29 UTC
Permalink
Post by Mark H Weaver
Post by Kei Kebreau
Post by Mark H Weaver
kkebreau pushed a commit to branch master
in repository guix.
commit 9e3a8ed0ebc5b4095f1b64d85fd56ce7fb636580
Date: Mon Dec 4 17:56:37 2017 -0500
gnu: python-pillow: Update to 4.3.0.
https://hydra.gnu.org/build/2400760
https://hydra.gnu.org/build/2400819
I restarted both of these, and they each failed again the second time.
The same builds (python-pillow and python2-pillow) also failed on armhf,
although as I write this they have not yet completed their second build
attempts.
The (new) failure on armhf also appears to be consistent, although
different from the i686 failure. Instead of a failed test, the build
https://hydra.gnu.org/build/2400971
--- 57 tests passed.
...SSSSS.......S..SS.....................................................SS...................................SSSSS..........................................................................................................................S....S.S.................S....................S.............................................................................S.building of `/gnu/store/cjinxvws27bwdwn7n2fab3k10had6y2p-python2-pillow-4.3.0.drv' timed out after 3600 seconds of silence
@ build-failed /gnu/store/cjinxvws27bwdwn7n2fab3k10had6y2p-python2-pillow-4.3.0.drv - timeout
Post by Kei Kebreau
I'm investigating now. Thanks for the heads up.
Thanks!
Mark
The i686 failure is possibly related to this bug:
https://github.com/python-pillow/Pillow/issues/2758

I've attempted to use the upstream patch, but it involves some GIT
binaries which aren't supported by GNU patch. Among other hackish
options, temporarily upgrading to the appropriate upstream commit may
fix this for now.

As for the armhf build, I'm still not sure what is going on.
Leo Famulari
2017-12-25 00:03:10 UTC
Permalink
Post by Kei Kebreau
I've attempted to use the upstream patch, but it involves some GIT
binaries which aren't supported by GNU patch. Among other hackish
options, temporarily upgrading to the appropriate upstream commit may
fix this for now.
Does it work if you pass '--binary' to patch-flags?
Kei Kebreau
2017-12-25 16:44:14 UTC
Permalink
Post by Leo Famulari
Post by Kei Kebreau
I've attempted to use the upstream patch, but it involves some GIT
binaries which aren't supported by GNU patch. Among other hackish
options, temporarily upgrading to the appropriate upstream commit may
fix this for now.
Does it work if you pass '--binary' to patch-flags?
It does not. Maybe I could use 'diff --binary' to patch the files into
existence?
Kei Kebreau
2017-12-25 17:09:23 UTC
Permalink
Post by Kei Kebreau
Post by Leo Famulari
Post by Kei Kebreau
I've attempted to use the upstream patch, but it involves some GIT
binaries which aren't supported by GNU patch. Among other hackish
options, temporarily upgrading to the appropriate upstream commit may
fix this for now.
Does it work if you pass '--binary' to patch-flags?
It does not. Maybe I could use 'diff --binary' to patch the files into
existence?
Using 'diff --text' on the non-text files seems to work. The patch file
is encoded as 'raw-text' according to Emacs. Please take a look at our
bug #29849.

Loading...