Open sthibaul opened 11 months ago
(this is preventing liblouis 3.28.0 from entering Debian Trixie)
See also #101
Failing to fix this, I will have to ask for dropping it from Debian...
If something needs to be dropped, can it please be liblouisutdml?
I asked for that, yes.
Should be fixed in cdfdc36e67abf4c1680e142817239eb4aa8e5b71. @sthibaul do you need a proper release?
No, I can cherry-pick.
I would however not really call xfail additions that a "fix" :)
I agree :)
@egli Perhaps you can at least add comments to the xfails pointing to issue https://github.com/liblouis/liblouisutdml/pull/101, so that the person who wants to fix it has a clue what could be the problem?
Oh, I had totally forgotten about #101. Will add comments.
I think this issue is obsolete as we are at 3.30.0 now. It should be fixed by #107 anyway
Did #107 also remove the xfails?
Did #107 also remove the xfails?
I don't think so
@hammera can you have a look at this? Maybe the fix is similar to #107
Ok Chris, I will looking this in weekend. A short help please: How can I easyest founding what virtual dot change affecting this tests with linked in #101 with final temporary we marked xfail? Need me analyze the expected.txt files and compare the dutch table change affected subtable files the escape sequences? Perhaps again need doing a simple modification the wiskunde.ctb table because this table are not folluw up yet all changes possible with added the 3.28, 3.29.0 releases, the 3.30 version affected testcases are already we together fixed.
Attila
Okay, I reverted the Liblouis 3.28.0 xfail markup related patch temporary my local repository (not pushed yet), now following tests are failing in Liblouisutdml master branch the Liblouis 3.30.0 component (perhaps similar the issue with we previous fixed in #107 related PR, but this tests are affected perhaps different virtual dot combination change with dutch braille related or a subtable change related with contains new virtual dots, not the 456a dot combination change):
Looks the verbose failed tests results output this six failing tests:
FAIL: test_mathml_woluwe/test_080
=================================
warning: lou_getDataPath is deprecated.
warning: lou_getDataPath is deprecated.
Testing MathML to Woluwe braille.Diff:
--- /usr/src/liblouisutdml/valami/liblouisutdml/tests/test_mathml_woluwe/test_080.test/expected.txt 2022-12-12 12:31:51.736096753 +0100
+++ /tmp/fooEqTSPV/output.txt 2024-06-07 13:47:48.424440919 +0200
@@ -1 +1 @@
-.IS #G20"-#C24 KLEINER DAN #C17,5
+.IS G\20"-C\24 KLEINER DAN C\17,5
FAIL test_mathml_woluwe/test_080.test (exit status: 1)
FAIL: test_mathml_woluwe/test_099
=================================
warning: lou_getDataPath is deprecated.
warning: lou_getDataPath is deprecated.
Testing MathML to Woluwe braille.Diff:
--- /usr/src/liblouisutdml/valami/liblouisutdml/tests/test_mathml_woluwe/test_099.test/expected.txt 2022-12-12 12:31:51.736096753 +0100
+++ /tmp/fooGRiyV7/output.txt 2024-06-07 13:47:48.848442815 +0200
@@ -1 +1 @@
-#/N:X*M:*Q:/P:
+*Q:/P:
FAIL test_mathml_woluwe/test_099.test (exit status: 1)
FAIL: test_mathml_woluwe/test_106
=================================
warning: lou_getDataPath is deprecated.
warning: lou_getDataPath is deprecated.
Testing MathML to Woluwe braille.Diff:
--- /usr/src/liblouisutdml/valami/liblouisutdml/tests/test_mathml_woluwe/test_106.test/expected.txt 2022-12-12 12:31:51.736096753 +0100
+++ /tmp/foomobFp8/output.txt 2024-06-07 13:47:49.016443562 +0200
@@ -1 +1 @@
-X/#A2Y:
+X/A\2Y:
FAIL test_mathml_woluwe/test_106.test (exit status: 1)
FAIL: test_mathml_woluwe/test_107
=================================
warning: lou_getDataPath is deprecated.
warning: lou_getDataPath is deprecated.
Testing MathML to Woluwe braille.Diff:
--- /usr/src/liblouisutdml/valami/liblouisutdml/tests/test_mathml_woluwe/test_107.test/expected.txt 2022-12-12 12:31:51.736096753 +0100
+++ /tmp/foo7fTkcH/output.txt 2024-06-07 13:47:49.040443672 +0200
@@ -1 +1 @@
-X/#A2:Y
+X/A\2:Y
FAIL test_mathml_woluwe/test_107.test (exit status: 1)
FAIL: test_mathml_woluwe/test_113
=================================
warning: lou_getDataPath is deprecated.
warning: lou_getDataPath is deprecated.
Testing MathML to Woluwe braille.Diff:
--- /usr/src/liblouisutdml/valami/liblouisutdml/tests/test_mathml_woluwe/test_113.test/expected.txt 2022-12-12 12:31:51.740096764 +0100
+++ /tmp/foozJJ7Lq/output.txt 2024-06-07 13:47:49.172444259 +0200
@@ -1 +1 @@
-%#C4"-#B:
+%C\4"-#B:
FAIL test_mathml_woluwe/test_113.test (exit status: 1)
FAIL: test_mathml_woluwe/test_114
=================================
warning: lou_getDataPath is deprecated.
warning: lou_getDataPath is deprecated.
Testing MathML to Woluwe braille.Diff:
--- /usr/src/liblouisutdml/valami/liblouisutdml/tests/test_mathml_woluwe/test_114.test/expected.txt 2022-12-12 12:31:51.740096764 +0100
+++ /tmp/fooUMnw1H/output.txt 2024-06-07 13:47:49.192444349 +0200
@@ -1 +1 @@
-%#C4:"-#B
+%C\4:"-#B
FAIL test_mathml_woluwe/test_114.test (exit status: 1)
It's almost like looking for a needle in a haystack, because the pull request related to the Dutch virtual score change changed a lot of places in individual sub-tables, display files, etc. :-):-) The from 4569 dot replacement to 456a changed the point combination, with which e.g. the previous faulty tests were corrected in the Liblouisutdml lbu_files/wiskunde.ctb table, I founded now them in the following tables and sub-tables in the Liblouis tables directory, but perhaps not this is the right subtable files and display files with real affecting the final fixed previous testcases in Liblouis side:
nl-BE.dis:display _ 456a _
nl-print.dis:display _ 456a
For example, I seeing this now my local repo failing tests changed some time only one character, but in Liblouis what file I founding the proper dot combination with I need replace the lbu_files/wiskunde.ctb file? :-):-)
Bert, a question only: Longer time period not better to putting back wiskunde.ctb table and wiskunde_edit.ctb table the Liblouis tables directory to provide always syncing new changes the wiskunde.ctb and wiskunde_edit.ctb table too if the dutch braille again changed the virtual dots? Unfortunately, if the board remains in Liblouisutdml, we are always running after the events, as a new Liblouis version comes out and the system of virtual points changes in Dutch Braille. The problem is not that the standard requires a change or that there is a better idea, it's just that the Dutch virtual braille points and the wiskunde.ctb, wiskunde_edit.ctb tables, which are used by the voluve braille test system, should be synchronized at the same time. Certain test parts cannot be transferred, e.g. to the Yaml system somehow from the tests of the math_voluve system? Or does this require xml processing in the same way as for the Nemeth and Nemeth_Mathcat tests too? If I asked silly questions, or if the proposal to take the math tables back to Liblouis is not easible, please forgive me, I am not an expert at all in Dutch Braille and the math_voluve Braille system. :-):-)
Sorry the long comment, I was just thinking while trying to compare the test results and take a note my thoughts, to see if they are useful at the idea level. :-):-)
Attila
(no, 3.31.0 doesn't bring more failures, it's just the remainder of 3.30.0 and previous)
Oh, I forgot these test cases real fixing. The affected test case files if I reverting proper xfail parts what the right output, based with actual nl-be.dis and nl-print.dis with newset Liblouis 3.310 and 3.31 versions? If I known right, nl-print.dis file changed only in Liblouis 3.31 version to Bert added the 4567 simbol.
So, if anybody known the good expected output with voluve Braille related with need produce the lbu_files/wiskunde.ctb table (based if I described perhaps good a longer comment if I understanding right the voluwe tests), please anybody type a comment the proper testcase file name, and the expected braille output.
Attila
I tryed looking tests/tests_math_volume/test_080.test/expected.txt output back-translation with lou_trace nl-unicode.dis,../../lbu_files/wiskunde.ctb,braille-patterns.cti -b command. This rules are executed with backward direction (I don't no interesting or not the backward direction when we examining the five not fixed yet xfail tests):
.IS #G20"-#C24 KLEINER DAN #C17,5
.\24F/\234F/@0@0\1245F/20=-@0\14F/24@0\13F/\123F/\15F/\24F/\1345F/\15F/\1235F/@0\145F/\1F/\1345F/@0@0\14F/17,5
1. math . 236
2. exactdots @0 0
3. exactdots @0 0
4. digit 2 12
5. digit 0 245
6. math = 2356
7. punctuation - 36
8. exactdots @0 0
9. digit 2 12
10. digit 4 145
11. exactdots @0 0
12. exactdots @0 0
13. exactdots @0 0
14. exactdots @0 0
15. digit 1 1
16. digit 7 1245
17. midnum , 2
18. digit 5 15
The expected output is following (I don't no what the differences with newest dot output in Liblouis 3.28.0 and newer version the dutch dots or virtual dots related, similar the changed 456a dot with already fixed the lbu_files/wiskunde.ctb):
.IS #G20"-#C24 KLEINER DAN #C17,5
Bert, Are there any other best practices for reliably debugging Voluve Braille tests if someone is completely unfamiliar with the Voluve Braille system, similar with me?
Attila
Hello, Since the release of liblouis 3.28.0, liblouisutdml's testsuite fails:
Samuel