Re: libidn2-2.3.0, Solaris 11.3 and one test failure

[Tim Rühsen]

Also possible:

some unwanted translation before the test.

Here I see
##########B#à.\u05D0\u0308#à.א ̈#xn--0ca.xn--ssa73l##
OK
OK

In your log I see
##########B#.\u05D0\u0308#.א̈#xn--0ca.xn--ssa73l##
OK
Failed: _check_toASCII(.א̈) -> .xn--ssa73l (expected xn--0ca.xn--ssa73l)
439040

Please see that the character 'à' as the first label is completely
missing in your log. Without it, the result is correct - but of course
something else is expected (the expected result is hard-coded) and thus
the failure.

Does the compiler drop that character ? You could possibly check that
with 'strings' or some hex output of the executable.

Regards, Tim

On 4/2/20 5:02 PM, Tim Rühsen wrote:
> Hi Jeff,
>
> I remember seeing errors on Solaris. Iconv and charsets did not properly
> work or were not properly implemented on Solaris.
>
> The two cited lines don't mean anything - some failures are on purpose
> and expected. On Debian Linux I get errno 84 for those (Invalid or
> incomplete multibyte or wide character).
>
> What does 88 on Solaris mean ?
>
> Regards, Tim
>
> On 4/2/20 4:08 PM, Jeffrey Walton wrote:
> > Hi Everyone/Tim,
> >
> > It looks like libidn2 is catching one self test failure:
> >
> > PASS: test-punycode
> > FAIL: test-lookup
> > PASS: test-register
> > PASS: test-strerror
> > PASS: test-tounicode
> > ...
> > # TOTAL: 5
> > # PASS:  4
> > # SKIP:  0
> > # XFAIL: 0
> > # FAIL:  1
> > # XPASS: 0
> > # ERROR: 0
> >
> > I see the failures but nothing jumps out at me. It all looks Greek
> > (Chinese) to me:
> >
> > u8_to_u32(≮𝟎．謖SS�) failed (88)
> > u8_to_u32(≮𝟎．謖ss�) failed (88)
> >
> > Attached is test-suite.log.