Remove obsolete cfgs and bump Pillow to 3.0.0

[pbauer]

No description provided.

[pbauer]

test: http://jenkins.plone.org/job/pull-request-5.0/434/

[gforcada]

Seems P.PortalTransforms fails due to a missing dependency on libtiff, I will look into it

[gforcada]

Trying again: http://jenkins.plone.org/job/pull-request-5.0/435

[mauritsvanrees]

Jenkins says:

AssertionError: [II*� ����]

!=

[II*� ����]

Locally it is very slightly clearer, with a x04 versus x03 when comparing the first twenty bytes of expected and actual output:

(Pdb) first
'II*\x00\x08\x00\x00\x00 \x00\x00\x01\x04\x00\x01\x00\x00\x00\x18\x00'
(Pdb) second
'II*\x00\x08\x00\x00\x00 \x00\x00\x01\x03\x00\x01\x00\x00\x00\x18\x00'
(Pdb) first == second
False

This is with a small Plone logo. Visually it looks the same to me.

[jensens]

The part/changed byte is in the image file directory (IFD), see http://www.fileformat.info/format/tiff/corion.htm

looks like this is related to Pillow Changes for 3.0.0
https://github.com/python-pillow/Pillow/blob/master/CHANGES.rst
"Tiff ImageFileDirectory rewrite #1419 [anntzer, wiredfool, homm]"

and looks like this was fixed for 3.1.0 unrelased (not sure, but sound like):
Flatten tiff metadata value SAMPLEFORMAT to initial value, fixes #1466 [wiredfool]

[pbauer]

Then let's wait for 3.1.0 or 3.0.1 and merge it when we don't have to fix a test. It's not urgent.

[jensens]

digged deeper, the fix for 3.1.0 seems unrelated python-pillow/Pillow#1466

digging further this is the major change python-pillow/Pillow#1419 I suspect as source of our problems.

[jensens]

current test failure seems to be a test setup problem, see plone/jenkins.plone.org#141

[gforcada]

@jensens not a test setup rather a junit parser error as the error is still there.

Whenever in doubt just look at the full console log: http://jenkins.plone.org/job/pull-request-5.0/695/console

[jensens]

well, but how can a pillow update break the test output. strange. maybe the test is crap.

[gforcada]

Because the output is binary :-)
El dia 16/01/2016 17:01, "Jens W. Klein" notifications@github.com va
escriure:

well, but how can a pillow update break the test output. strange. maybe
the test is crap.

—
Reply to this email directly or view it on GitHub
#124 (comment)
.

[jensens]

So I fixed the output in PortalTransforms to be able to analyse the problem. See plone/Products.PortalTransforms#11

The TIFF header (first 8 Bytes) are ok.

In the following (in test output truncated) IFD headers are changed: the byte at position 13 is a 4 instead of a 3. This byte indicates the value-type (3 is a word, 4 a dword) of tag 0100 (Image width).

Further reading: http://www.fileformat.info/format/tiff/corion.htm http://paulbourke.net/dataformats/tiff/
I do not see any problem here. I'am not sure what the reason behind this is - mabe there are images wider than short integers max are used - but any code reading IFD headers must support this.

I propose to change PortalTransforms so it

• before branch away a 2.1.x branch to be used in Plone 4.3,
• bump version on minor level (2.2.0),
• depend on Pillow>=3.0.1
• uses a expected output test file with a 4 in the header.

Opinions?

[thet]

+1

[mauritsvanrees]

+1

[jensens]

done.