gnustep base test failure

[thomas kaeding]

One test failed (-ve number format).

Tests/test.log attached as "check.log"

gnustep-base-1.25.0
gnustep-make-2.7.0
GNU/linux with glibc 2.26 and kernel 4.14.2 and GCC 7.3.0

I did NOT subscribe to the list, so if there is a patch, I would be
grateful to receive an email.

[Richard Frith-Macdonald]

Post by thomas kaeding
One test failed (-ve number format).
Tests/test.log attached as "check.log"
gnustep-base-1.25.0
gnustep-make-2.7.0
GNU/linux with glibc 2.26 and kernel 4.14.2 and GCC 7.3.0
I did NOT subscribe to the list, so if there is a patch, I would be
grateful to receive an email.

I suspect this is nothing to worry about ... not a bug, but rather a failing of that particular testcase (checking the format of a negative currenty for Brazilian Portugese).
The testcase supports two variants, because different version of the ICU library on different systems produce different results.
It seems likelly that the version of ICU on your system is using a third variant of format for negative currency values, one the testcase does not allow for.
I have changed the testcase in git to use a fixed negative currency format, rather than the one ICU supplies for the pt_BR locale, in order to avoid that variability.

[Yavor Doganov]

Post by Richard Frith-Macdonald
I have changed the testcase in git to use a fixed negative currency
format, rather than the one ICU supplies for the pt_BR locale, in
order to avoid that variability.

Now it fails for me with ICU 57.1 (str is _R$1.235) and 60.2 (str is
_R$<A0>1.235, see Bug#53035).

[Richard Frith-Macdonald]

Post by Yavor Doganov

Post by Richard Frith-Macdonald
I have changed the testcase in git to use a fixed negative currency
format, rather than the one ICU supplies for the pt_BR locale, in
order to avoid that variability.

Now it fails for me with ICU 57.1 (str is _R$1.235) and 60.2 (str is
_R$<A0>1.235, see Bug#53035).

Sorry, I failed to notice bug #53035
My attempted fix was wrong, so I applied your patch. Thanks.