home *** CD-ROM | disk | FTP | other *** search
- Path: sparky!uunet!olivea!apple!goofy!mumbo.apple.com!gallant.apple.com!delos.apple.com!user
- From: shebs@apple.com (Stan Shebs)
- Newsgroups: comp.sys.mac.programmer
- Subject: Re: what is the best C compiler
- Message-ID: <shebs-050193104435@delos.apple.com>
- Date: 5 Jan 93 18:54:12 GMT
- References: <1992Dec19.165625.2597@ll.mit.edu> <bauer-221292105134@134.60.68.23> <keith-221292125608@kip-28.taligent.com> <1993Jan1.145255.24376@netcom.com>
- Sender: news@gallant.apple.com
- Followup-To: comp.sys.mac.programmer
- Organization: Apple Computer Inc.
- Lines: 29
-
- In article <1993Jan1.145255.24376@netcom.com>, jimlynch@netcom.com (Jim
- Lynch) wrote:
- > Keith Rollin sed:
- > GNU C lost a lot of ground because it was unable to convert normal
- > branches into short branches when appropriate.
- >
- > One of the beautiful parts of GNU c is that you can change this, and probably
- > Keith's other complaints as well... How? by editing the machine description
- > and recompiling.
-
- Actually, short branches can't be handled easily by GCC, because the
- compiler doesn't normally keep track of the lengths of instructions;
- instead, it has huge numbers of fprintf calls that build up the asm
- text in a single pass. So although GCC supports many wonderful
- features, shortening branch addresses is not one of them.
-
- Fortunately, the MPW Asm has an amusing quirk, which is that it
- reliably identifies short branches without being able to actually
- generate them. So, Keith put together some code a year ago that
- just collects the warning messages and bashes the asm code, then
- reruns the assembler. You can keep rerunning until the assembler
- no longer issues any warnings... Anyway, it's incorporated into
- a new release of MPW GCC 1.37.1 that I've been preparing over the
- past month, as an interim release until 2.x is available. Should
- be ready any day now! (yeah yeah, promises promises :-) )
-
- Stan Shebs
- Apple ATG System Software
- shebs@apple.com
-