Front page | perl.perl5.porters |
Postings from May 2001
Michael G Schwern <schwern@pobox.com> writes: >On Wed, May 30, 2001 at 02:26:04PM +0000, Nick Ing-Simmons wrote: >> Maybe - we are not sure or we would have done it a long time ago >> with a perl one-liner. Most are okay, but there are probably >> lingering spots where n_a is used as a scratch variable and value >> _is_ used. > >So if n_a isn't explicitly used in the code its safe to assume it can >go? If n_a does not occur in the code it isn't there to be removed - I don't understand. > > >> >PS What's the difference between SvPV and SvPVx? >> >> Read the source (i.e. sv.h) ... > >*cough*shouldbeinperlapi*cough* ;) Patches welcome - one reason it isn't is we are not proud of this mess... > >> SvPVx makes a copy of the SV * so that >> >> SvPVx(*p++,len) is "okay" >> whereas >> SvPV(*p++,len) isn't as the macro may do multiple ++s. > >Right, but why only for gcc? You stopped reading before you got to the others ? It is made "safe" 3 ways: 1. GCC special 2. Use of icky per-thread-global SV * 3. By being a real function not a macro. -- Nick Ing-Simmons who is looking for a new job see http://www.ni-s.u-net.com/Thread Previous