Front page | perl.perl5.porters |
Postings from June 2012
On 6/19/2012 2:56 PM, bulk 88 wrote: > ---------------------------------------- >> Date: Tue, 19 Jun 2012 14:44:31 -0700 >> From: rev.chip@gmail.com >> To: bulk88@hotmail.com >> CC: perl5-porters@perl.org >> Subject: Re: Time for a 64-bit perl API? >> >> On 6/19/2012 2:33 PM, bulk 88 wrote: >>> ---------------------------------------- >>>> Date: Tue, 19 Jun 2012 13:34:32 -0700 >>>> From: rev.chip@gmail.com >>>> To: jand@activestate.com >>>> CC: bulk88@hotmail.com; perl5-porters@perl.org >>>> Subject: Re: Time for a 64-bit perl API? >>>> >>>> Indeed. An error value for an unsigned type, in a properly portable >>>> program like Perl, would be ~(type)0 or (type)-1, not bare -1. >>> http://perl5.git.perl.org/perl.git/blob/HEAD:/uconfig.h#l2978 >>> http://perl5.git.perl.org/perl.git/blob/HEAD:/regexp.h#l184 >> Don't know when to stop, do you? >> These are cases that prove my point. >> >> RD_NODATA is a signed value; it is the error flag for the read() >> function, which returns ssize_t (or in Perl, SSize_t). >> RX_BUFF_IDX_POSTMATCH is a signed value; it is assigned to "register >> I32 paren". >> >> C'mon, keep going. This is fun. >> > I dont see a cast in that macro, do you? I don't see "unsigned" in those types, do you?Thread Previous | Thread Next