Re: make install using "-j"

[Paul Eggert]

"David Everly" <address@hidden> writes:

> I've also opened a bug on mkdir -p against ia64-hp-hpux11.23, and I
> want to confirm this bug (I'm not an expert on posix, especially
> relating to umask and mkdir).

I'm sure HP will thank you for your efforts....

> Here are the three tests, including the
> actual results and the expected results (please help me review that my
> expected results are correct):
>
> ## Test one:
>
> address@hidden:~/mkdirtest$ /usr/bin/umask 002

The test should invoke the plain 'umask' that is built into the shell.
'/usr/bin/umask 022' is a no-op, for the same reason that '/usr/bin/cd /'
is a no-op.  I'm afraid that you'll have to redo this test.

Test two is correct (i.e., you correctly demonstrate a bug).

> ### Test three:
>
> address@hidden:~/mkdirtest$ /usr/bin/umask
> 022
> address@hidden:~/mkdirtest$ /usr/bin/mkdir -m 775,u+wx -p -- foo/bar

This is less important, because the '775,u+wx' usage doesn't conform
to POSIX, and install-sh doesn't use it.  Since the behavior is
undefined in this case, it's not that big a deal what mkdir does
(though HP may want to fix the behavior anyway).  GNU 'mkdir -m
755,u+x' reports an error, which is a reasonable thing to do.