regex tests fail in no_unicode branch

[Daniel Turcanu]

Some regex tests fail in no_unicode branch when executed with ICU regex.
Those tests are:

1458:test/rbkt/zorba/string/Regex/regex_err10
1460:test/rbkt/zorba/string/Regex/regex_err12
1463:test/rbkt/zorba/string/Regex/regex_err15
1464:test/rbkt/zorba/string/Regex/regex_err16
1474:test/rbkt/zorba/string/Regex/regex_err25
1478:test/rbkt/zorba/string/Regex/regex_err7
1483:test/rbkt/zorba/string/Regex/regex_m11

I mark them as expected failures.

[Paul J. Lucas (paul-lucas)]

I don't understand. If NO_UNICODE=ON, then ICU's regex should *not* be used and *your* regex should be used.

[Daniel Turcanu (danielturcanu)]

The no_unicode branch can also be compiled with NO_UNICODE=OFF. This is the case I was testing.

[Paul J. Lucas (paul-lucas)]

OK, I will look into this. But this bug should not impede progress of finishing the no-Unicode work. This bug is a pre-existing bug and has nothing to do with the no-Unicode changes.

[Daniel Turcanu (danielturcanu)]

From my point of view, no unicode branch is finished.
At least on Windows pass all tests that pass in normal mode.
I filed this bug with low priority to mark those tests as expected failures. I don't expect the bug to be solved. Just try to see what you can do and then close it with Won't fix.

[Paul J. Lucas (paul-lucas)]

For now, tests regex_m11 and regex_err15 should be marked as "expected failure" only when ZORBA_NO_ICU=OFF. The relevant bug is bug #866874.

The tests regex_err7, regex_err12, and regex_err25 have been fixed.
The test regex_err10 has a new, specific bug opened for it: bug #974474.
The test regex_err16 has a new, specific bug opened for it: bug #974477.

Note that both the existing and new bugs are pre-existing failures and have nothing to do with the no-Unicode changes. Therefore, they should not stop the merge of the no_unicode branch.