System V/ANSI C: this is the wrong way to do this, do *not* use these.
Of course, good old groff IS using it and fails when creating the .ps
versions of clang man pages.
How do we fix this?
A) Stop groff from using P_tmpdir and default to "/usr/tmp".
B) Change stdio.h to "/usr/tmp".
C) Let "make world" create a symlink /var/tmp -> /usr/tmp.
D) Any combination of the above.
I personally lean towards A) + B). Since nbsd tells us that groff is
doing it wrong, A) is the right thing to do anyway. But there may be
more packages out there using P_tmpdir and they don't fail during
compile time, they fail at runtime if and when they need a temp file.
Opinions?
Jan
--
Anyone who trades liberty for security deserves neither
liberty nor security. -- Benjamin Franklin
My bad, I didn't fetch/pull for 2 days :)
Never mind then ...
Do mind! We install the nbsd_include/stdio.h in /usr/include now.
>>> Both /usr/src/include/stdio.h and /usr/src/nbsd_include/stdio.h define
>>> P_tmpdir as "/var/tmp", a directory that doesn't exist by default. The
>>> nbsd stdio.h has a comment that says:
>>
>> Actually, include/stdio.h specifies "/tmp" (see
>> http://git.minix3.org/?p=minix.git;a=blob;f=include/stdio.h;h=f8e355d77204eae1dd8bfe75101e299f67ceb46b;hb=HEAD).
>
> My bad, I didn't fetch/pull for 2 days :)
Actually it has since the constant was added a year ago.
Please note that I'm referring to the MINIX libc one (include/stdio.h),
the NetBSD libc (nbsd_include/stdio.h) one is indeed inconsistent.
With kind regards,
Erik