Hi Chris and others,
Some more about the yaml harness.
Changing the way begcaps/endcaps are handled in the UEB patches squashes a
number of of failures. See my e-mail from yesterday.
Backtranslation still isn't up to par, see hu-hu-g1_backward.yaml. An
interesting point here is that check_yaml reports 1 failure whereas the output
shows 2 failures. Bug?
Some other failures look suspiciously like errors in the test data, but I
haven't looked into this yet. For some examples, see the output of
fi_harness.yaml. The test for URLs is a good example. It looks fine in the
Python harness but not in the converted yaml.
Other failures were due to the tests using Unicode 0x2800 (empty braille cell)
instead of regular spaces. There's a PR on GitHub to correct these.
It would be nice to show the typeforms for failures that use them. Any thoughts
on how to do this in a pretty way? See for example
en-us-g2-end-emphasis_harness.yaml. It's not very clear why these are failing
now.
Could the conversion from Python harness to yaml be automated, maybe during
make check? I'm actively developing the Pyython harness tests for the Dutch
tables and would like to keep the yaml side of things up-to-date.
Davy
-----Oorspronkelijk bericht-----
Van: liblouis-liblouisxml-bounce@xxxxxxxxxxxxx
[mailto:liblouis-liblouisxml-bounce@xxxxxxxxxxxxx] Namens Christian Egli
Verzonden: woensdag 15 juli 2015 14:54
Aan: liblouis-liblouisxml <liblouis-liblouisxml@xxxxxxxxxxxxx>
Onderwerp: [liblouis-liblouisxml] Status ueb integration
Hi all
Just a heads-up on the status of the integration of the ueb changes into
liblouis main line:
From Mike Gray's repository I have pulled the latest changes into the
feature/ueb_update branch. On a separate branch (integration/ueb_merge) I
merged this with the yaml tests (feature/yaml_tests).
On feature/ueb_update `make check` passes with no failures (ueb_test_data.pl is
marked as expected failure).
On feature/yaml_tests the tests (including the yaml-based harness tests) pass.
On the merged branch, i.e. on integration/ueb_merge I get 14 failures.
The output is attached. I suspect that this confirms many of the regressions
that Davy has been talking about. In particular Davy was mentioning regressions
with Numbers. I suspect the following excerpt from the test log confirms this:
FAIL: ar-ar-g1_harness
======================
Input: '4.0'
Expected: '⠼⠙⠨⠚' (length 4)
Received: '⠼⠙⠨⠼⠚' (length 5)
Or from the Korean table:
FAIL: ko-2006-g2_harness
========================
./ko-2006-g2_harness.yaml:5 Failure
Input: '2000년'
Expected: '⠼⠃⠚⠚⠚ ⠉⠡' (length 8)
Received: '⠼⠃⠚⠚⠚⠉⠱⠒' (length 8)
Diff: Expected ' ' but received '⠉' in index 5
I will try to investigate this some more when I get to it.
Thanks
Christian
-----
Die Android-App für SBS-Hoerbuecher ist da! Jetzt kostenlos im Google Play
Store herunterladen.
(https://play.google.com/store/apps/details?id=nl.dedicon.sbsleser)
DISCLAIMER:
De informatie verzonden met dit e-mail bericht is uitsluitend bestemd voor de
geadresseerde. Indien u niet de beoogde geadresseerde bent, verzoeken wij u
vriendelijk dit aan de afzender te melden (of via:
info@xxxxxxxxxx<mailto:info@xxxxxxxxxx>) en het origineel en eventuele kopieën
te verwijderen.
The information sent in this e-mail is solely intended for the individual or
company to whom it is addressed. If you received this message in error, please
notify the sender immediately (or mail to
info@xxxxxxxxxx<mailto:info@xxxxxxxxxx>) and delete the original message and
possible copies.
