Front page | perl.perl5.porters |
Postings from February 2012
On Sat, Feb 11, 2012 at 9:42 PM, Ævar Arnfjörð Bjarmason
<avarab@gmail.com> wrote:
> I very much like where this patch is going, but IMO it needs some
> improvement:
>
> * You've changed it so that we now return the current and correct
> values when getting $<, $>, $( and $), but there's still a lot of
> places in the perl core where we're looking at PL_uid and pals,
> these should be changed to call PerlProc_getuid et al instead.
Yeah, that'd be a good idea. Though doing it properly may invite some
refactoring
> * Since the id swapping is only used by "PL_delaymagic &= ~DM_RUID;"
> and pp_sassign (and variants) we should just add new a new
> PL_delaymagic_uid variable that's only used by the set() magic and
> sassign().
>
> This variable should not be made public. Thus programs on the CPAN
> that expect to assign to PL_uid would fail and would need to be
> updated, but they wouldn't contain a logic error anymore (by
> expecting the core to read from PL_uid).
>
> * Code like the code in this ifdef in POSIX.xs can just go away:
>
> SysRet
> setuid(uid)
> Uid_t uid
> CLEANUP:
> #ifndef WIN32
> if (RETVAL >= 0) {
> PL_uid = getuid();
> PL_euid = geteuid();
> }
> #endif
>
> * It would also be informative to check how much of the CPAN is
> relying on PL_uid, but I don't think that should block this going
> in.
I took my approach because it's easiest. Existing code would
effectively become a no-op. In principle it's a good idea, but given
this may break stuff on CPAN, depends on #1 and we're approaching the
user-visible features deadline fast I'd rather postpone it until 5.17.
Leon
Thread Previous
|
Thread Next