Mail Archives: djgpp/1997/09/14/15:06:55
Ron Hiler <bndwgn AT pacbell DOT net> wrote:
>Avery Lee wrote:
>Only the
>> crappiest of 32-bit compilers (hmmm... Borland? :) ) would be beaten by a
>> 16-bit compiler in terms of the speed of a serious program.
>
>This comment is of some concern to me. I use Borland compilers, and am
>about to upgrade to the new version (either Builder or 5.0 (right?)).
>Is there some significant reason Borland is bad? Should I go with
>something else? (Please dont say DJGPP, I'm addicted to the IDE).
It was meant to be a joke, but at least with certain versions of their
compiler, Borland has been known to sacrifice code speed for compilation
speed. I've never found a bug in the optimized code generation of Watcom,
but I've found at least one serious bug in Turbo C++ 3.1, where a huge
pointer did not increment its segment pointer properly. This is a BIG bug,
and the only workaround I found was to use inline assembly to increment the
pointer.
Powerpack 32-bit programs are a special pain because they require Borland's
VxD to support paging under Windows, or else they refuse to run. This is a
real pain, and unnecessary under the Windows environment, where DPMI
programs are already paged. If you're planning on writing 32-bit DOS, I'd
say go to Watcom, and if you want to program Win32, go to MSVC++. I can't
see any reason for using Borland anymore.
But then again, this is only my opinion.
-- Avery Lee (Psilon AT concentric DOT net)
Try my freeware junk at http://www.concentric.net/~Psilon
(pages revised 8/6/97 - all programs include source)
* VGAPaint 386 V1.3/1.4b4: 32-bit paint program
* VETools 1.21: 2:1 compression for Watcom executables
* IPXfer5 1.6a: easy file transfers over networks at 450k/s
* Playwav 1.2: PC speaker sound for older computers
* Stellaryx/PC 1.1: Shoot-em-up from the Amiga, ported to the PC
* LMPStat 1.0: Pages of statistics for DOOM ][ .LMP files
- Raw text -