Author: Dann Corbit
Date: 17:51:00 06/18/01
Go up one level in this thread
On June 18, 2001 at 20:14:13, David Rasmussen wrote: >On June 18, 2001 at 18:16:53, Gian-Carlo Pascutto wrote: > >>On June 18, 2001 at 18:02:33, Dann Corbit wrote: >> >>>On June 18, 2001 at 12:38:19, Dan Andersson wrote: >>> >>>>http://www.gnu.org/press/2001-06-18-GCC.html >>> >>>It seems to be vapor-ware of some sort. >>> >>>I followed every link on the planet, and can't find a downloadble version at >any of the ftp sites. >> >>Slashdot effect or something. It should have been propagated to >>the mirrors before the announcement came out, but it's not >>unlikely those are out too. >> >>It'll break a lot of code. They may argue the code that breaks >>was just broken already, but not being backwards-compatible sucks >>anyway. >> > >I disagree completely. It rules big time. If people want backwards >compatibility, they can use old stuff. Depends on the level of pain. Suppose that we can make a compiler that is four times faster, but there is no such thing as malloc(), printf(), or fseek(). Will you actually use it? If there is minimal effort needed for a recompile, it might be a good idea. The necessary evil of backwards compatibility has been addressed (for instance) by the ANSI/ISO C committees, and I think their conclusion (some is needed) is the right one. >Making a clean break makes it possible to >make a more well-designed system without regard for old odd idiomatic bugs and >features of the old version. I think following the standard better is a good idea, generally. Not sure what you mean by "a clean break" since they are definitely not going to do a rewrite from scratch. >Intel should do the same. Make a clean break. If they do, I hope they don't break anything or slow it down. They have the best compiler on the planet [on average -- YMMV].
This page took 0 seconds to execute
Last modified: Thu, 15 Apr 21 08:11:13 -0700
Current Computer Chess Club Forums at Talkchess. This site by Sean Mintz.