Re: POSIX replament functions on HP-NonStop

[Bruno Haible]

Joachim Schmitz wrote:
> OK, thanks.
> 
> The alignof problem is still not solved, but I managed to run a 'make check'
> and here's where it fails: 

Good! At least the entire testdir now compiled without errors.

> FAIL: test-alignof                                   <<<< ignore for now

I'm out of ideas for investigating this one.

> /usr/local/Floss/gnulib/testdir-posix/gltests/test-mbsrtowcs.c:118: assertion 
> failed
> ./test-mbsrtowcs1.sh[14]: 326041862 Abort
> FAIL: test-mbsrtowcs1.sh

Needs to be investigated with a debugger or with printf statements.

> /usr/local/Floss/gnulib/testdir-posix/gltests/test-nanosleep.c:76: assertion 
> failed
> /bin/sh: 477036627 Abort
> FAIL: test-nanosleep

Likewise.

> /usr/local/Floss/gnulib/testdir-posix/gltests/test-rename.h:178: assertion 
> failed
> /bin/sh: 611254355 Abort
> FAIL: test-rename
> /usr/local/Floss/gnulib/testdir-posix/gltests/test-rename.h:178: assertion 
> failed
> /bin/sh: 661586003 Abort
> FAIL: test-renameat

Likewise.

> Unconnected socket test... passed
> Connected sockets test... passed
> General socket test with fork... passed
> Pipe test... failed (expecting POLLOUT | POLLWRNORM before writing)
>   failed (expecting POLLOUT before writing)
>   failed (expecting POLLWRNORM before writing)
>   failed (expecting POLLIN | POLLRDNORM after writing)
>   failed (expecting POLLIN after writing)
>   failed (expecting POLLRDNORM after writing)
> FAIL: test-poll

This test fails also for many other platforms.

> Unconnected socket test... passed
> Connected sockets test... passed
> General socket test with fork... passed
> Pipe test... failed (expecting writability before writing)
>   failed (expecting writability before writing)
> FAIL: test-select

select() on pipes appears to be broken. I have no idea whether a workaround
is possible.

> /usr/local/Floss/gnulib/testdir-posix/gltests/test-sigaction.c:81: assertion 
> failed
> /bin/sh: 225378387 Abort
> FAIL: test-sigaction

This one is probably not important.

> =======================
> 8 of 314 tests failed
> (27 tests were not run)
> =======================

That's a pretty good result, given that you only started to port gnulib two or
three months ago.

Bruno