On 9/22/2011 1:46 AM, Nick Keighley wrote:
> On Sep 22, 8:12 am, Lucien Coffe<l...@nospam.invalid> wrote:
>> Joshua Maurice wrote :
>>
>>> Is this a poor attempt at a troll?
>>
>> Either that or this guy has a time machine and comes from the 80's.
>
> HLLs existed even in the 80s
yes, but IIRC, but back in the days of MS-DOS, it was more common to
write much of ones' code in assembler.
by the time dos-extenders and similar were getting popular, most
development had shifted to C and C++ (and since then ASM has taken on an
increasingly smaller role in most apps).
Unix style OS's (namely Linux) didn't really start gaining any real
popularity on PC-style hardware AFAIK until around the mid/late 90s.
some of my project though does involve using ASM, but nearly all of it
is dynamically generated (an assembler exists in-program, and used for
assembling code at runtime). a merit of ASM though is that it is very
capable, and compiles/assembles very quickly. it currently supports x86
and x86-64, ARM and Thumb (ARM/Thumb support still needs work though).
its big drawback (fairly obvious) is that it is specific to each
combination of CPU/mode/OS/... meaning that it is not generally used for
general-purpose logic code (everything that works fine in C or C++ is
left in said languages).
mostly, it is used for specific features which can't be implemented
(effectively, or at all) at the same level as C or C++, which means
mostly reflection-related operations, and for JIT-related purposes (it
is IMO much nicer, and easier to debug, to produce ASM as output from a
JIT, rather than going directly to machine-code, and generally the
assembler is plenty fast enough that its own overhead generally doesn't
matter so much).
or such...
|