Re: Current test version of Chess 0.5
- From: frank@xxxxxxxx
- Date: 24 Jun 2005 18:55:28 GMT
I Forget <No@xxxxxxxxxxx> wrote:
> <frank@xxxxxxxx>
>
>>> Maybe it's time for a new standard. One that's actually updated for
>>> the real world...[sigh]
>>
>> It's called ISO 10206 and was released in 1990. D'oh!
>
> That only does some stuff.
>
> Other, such as the REAL problem is still very much present.
And which other dialect of Pascal does this? Sure, the Borland
manuals describe the layout of real types in detail. But their
compilers are implemented only on one CPU, so for the most part they
only had to copy the hardware description (except for their
self-made 6 byte type, which today is much less efficient and
therefore probably hardly used anymore except for historical
reasons).
Do you think if Borland were to port their compilers to another CPU
with different FP hardware, they'd emulate the IA32 types? I doubt
it, actually I'd hope not -- for performance this would probably be
quite a stupid thing to do.
Likewise, you can pick any other compiler (ISO or not) on a given
platform, verify that its FP sizes match your requirements, and
pretend this compiler guaranteed the FP range you need.
Fortunately, most FP hardware these days is IEEE compatible. But
this applies to any language dialect. When running on an IEEE
platform, you won't have much problems, when not, then good luck,
with any dialect.
Or what do you think would happen if the standards prescribed a
minimum size?
a) The size would be conservative enough so every wide-spread
platform supports it directly. Every implementor would verify
that it does and go ahead. You get the same effect now by
checking the size at program start (as has been described here).
b) The size would be a serious obstacle. I suppose few implementors
would bother to write an expensive and inefficient emulation.
Rather, they'd either omit this part of the standard or simply
avoid supporting platforms that don't support the size naturally,
possibly by a check a program start-time. Again, you can achieve
the same that by a self-made check. Actually, the latter is
preferable, since you can check for exactly the size you need,
rather than some arbitrary choice put in a standard.
Frank
--
Frank Heckenbach, frank@xxxxxxxx, http://fjf.gnu.de/
GnuPG and PGP keys: http://fjf.gnu.de/plan (7977168E)
Pascal code, BP CRT bugfix: http://fjf.gnu.de/programs.html
Free GNU Pascal Compiler: http://www.gnu-pascal.de/
.
- Prev by Date: Re: Current test version of Chess 0.5
- Next by Date: Re: Current test version of Chess 0.5
- Previous by thread: Re: Current test version of Chess 0.5
- Next by thread: Re: Current test version of Chess 0.5
- Index(es):
Relevant Pages
|
