David Scriven <[EMAIL PROTECTED]> wrote: > I stand corrected - the word I should have used was > 'suggested' as in "you will find a recent update > at this site". Whatever the case, gcc 2.96-85 DOES > have bugs (I have a large amount of code that works > perfectly on other compilers to attest to this) Does > Red Hat 'recommend' the Rawhide 2.96-99 rpm > as a solution?
Is it safe to assume that you have filed bug reports on bugzilla on each of the bugs which you mention that you have found? FWIW, as others have mentioned, MPlayer 0.50 compiled (and runs) just fine for me using gcc-2.96-85 -- after removing the config nonsense of course... One rather important point here is that successful compilation (and even execution) using this or that compiler does __NOT__ indicate that the code is correct. I have written significant amounts of code -- both professionally and personally -- which was not strictly correct, yet which compiled and executed the way I wanted it to (i.e., "correctly" in the context in which I was using it). My code simply took advantage of non-standard features of the compiler(s) I was using at the time. When I first tried 2.96-(earlier release) I was really irritated because a medium-sized project which I had just finished at the office would not compile using it. I tried 3 other compilers on 3 OS's and they all compiled it without a whimper (OK, warnings, but no errors). Then I smugly went back and more thoroughly examined what I had done. It was embarrassing but I had relied on some less-than-ansi-conformant tricks which happened to compile on those other compilers (one was an older gcc). That did not, however, make my code either ANSI-conformant or "correct" (whatever "correct" means)... Clearly gcc-2.96 is more strict about the standard (and it doesn't appear that the interpretation of the standard is really the subject here). And I do not think it reasonable to call insisting on conformance to the standard a bug, even though it makes more work for me. I would probably like to see an option added which enables compatability with the previous release(s?) of gcc, but (having written other compilers) I suspect that the addition of such an option would be a GoodThing (tm) since it could easily open up so many other potential areas for bugs. Yes, I also found one bug in the 2.96 releases a while back -- and I suspect that it actually was a bug. But I was never able to isolate the it, and after a few days trying I just re-coded the section which appeared to trigger it, and the run-time bug disappeared. Bill Austin [EMAIL PROTECTED] _______________________________________________ Seawolf-list mailing list [EMAIL PROTECTED] https://listman.redhat.com/mailman/listinfo/seawolf-list
