Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Compiler bug (OT)

Author: Miguel A. Ballicora

Date: 13:27:35 12/08/01

Go up one level in this thread


On December 08, 2001 at 11:33:14, David Rasmussen wrote:

>On December 08, 2001 at 10:45:55, Miguel A. Ballicora wrote:
>
>>On December 08, 2001 at 06:03:14, David Rasmussen wrote:
>>
>>>On December 08, 2001 at 04:16:46, Lex Loep wrote:
>>>
>>>>Altough a bit off topic I still post it as many here use the MSVC compiler.
>>>>This bug makes me feel a bit uneasy as this is just a simple constuction
>>>>and makes you wander how many more of these faults are in there.
>>>>
>>>>Lex
>>>>
>>>
>>>There are no compilers without bugs. I have found a bunch in various versions of
>>>gcc myself. In fact, I have discovered fewer compiler bugs in MSVC than in gcc,
>>>but that may just be a coincidence. On the other hand, neither of those two
>>>compilers are very standard compliant, and that may be viewed as a bug in
>>>itself. The gcc 3.0.x branch is far more standard compliant than older gcc's,
>>
>>gcc looks very compliant if you use the proper switches. IIRC you have to use
>>the switch -pedantic. What do you mean exactly by "not very standar compliant"?
>>You refer to the extensions they have?
>>
>
>-pedantic only turns on certain warnings. -ansi turns on some errors and
>warnings for stuff that isn't ANSI/ISO compliant. But I'm afraid that's not
>enough to be standard compliant :) . I am not talking about extensions either. I
>am talking about things that gcc does that contradicts the standard
>(streambuf::in_avail() in gcc 3.0.x branch for example), and stuff that gcc
>doesn't do at all, although it is in the standard (loads of template code in all
>versions, stringstream missing in 2.95.x branch). gcc's ANSI/ISO C++ compliance
>is known not to be complete at all. So are most other compilers, except maybe
>for Comeau C++ and KAI C++, and others using the same frontend. The biggest
>problems are with templates. gcc has lousy template support (as does MSVC and
>Borland C++). But there are lots and lots of small and not so small things in
>gcc that are not compliant to the standard. Many programmers don't know this, as
>they themselves do not write compliant code. They write code that works on the
>compiler that they are most used to. To be fair, gcc 3.1 branch looks promising.
>3.0.x had better compliance than 2.95.x, and 3.1 will have even more compliance.


Ok, I thought you were talking about ANSI C and not C++

Miguel



>
>/David



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.