https://bugs.exim.org/show_bug.cgi?id=2916
--- Comment #9 from Jeremy Harris <jgh146exb@???> ---
(In reply to Martin Kealey from comment #5)
> 3. check for the existence of a sentinel but then exclude it (ret=dir).
Is that changing the functionality of ret=dir? If so, back-compatitibility
argues against it.
> was trivial to include all the other standard S_IFxxx
> inode types, and they could be accommodated without runtime overhead.
All very well, but don't forget the ongoing maintenance costs of extra
cruft. Especially if it's of little use. I'm not rejecting it here, only
making the point.
(comment #6)
> the %#x format was defined by C89
... for a C library routine. You're not using one.
(comment #7)
> The UNUSED macro is to silence the compiler warnings
Chasing compiler warnings is pointless. They are endlessly proliferating.
Yes, we've built up cruft over the years in trying to do so.
If you're annoyed by your particular compiler and version of compiler being
verbose, turn that warning off using commandline options.
(comment #8)
> So I would suggest that the onus should be reversed
No. Change in behaviour, on those platforms.
--
You are receiving this mail because:
You are on the CC list for the bug.
