Re: Standard Library Interest?

["Robert I. Eachus" <rieachus@comcast.net>]

Marin David Condic wrote:
> Well, as an "end user" in good standing, I'd say it did affect me. I 
> wanted portable source code I could take unchanged from one compiler to 
> another and I couldn't get there. (Not initially, at least) So there was 
> an impact. If I was maintaining code for two different compilers (not at 
> all unheard of) I'd have to go through gyrations of configuration 
> management to deal with it in some manner.

True, but not realistic.  Having been there and done that, there were 
two situations that arose.  One, you had two compilers for the same 
hardware, and it was almost trivial to pick one version of the math 
libraries and treat it as part of your source code.

But the much more frequent case was where I was targeting different 
hardware.  Then choosing the "right" version of a library was 
complicated.  For example, I had three bodies of the math libraries for 
the Amiga: 68000/68010, 68020/68030 with 68881/68882(IEEE), and 
68020/68030 without math chip.  It would have been nice to be able to 
build a single library that did the necessary tests when initialized and 
used the "right" routines automatically, but that didn't work.  The 
floating-point data types and calling sequences were sufficiently 
different that you really had to compile for the floating-point model 
you were using.

It was no different for other languages, you had to choose an 
implementation that your machine supported, and it would run much slower 
than it had to if you choose the wrong version.  (Extremely slowly if 
you generated 68881 calls on a machine that didn't have a math chip.)

> Its an issue that can safely be put to bed now, but the lesson ought to 
> be out there for future similar circumstances. Not everybody's political 
> needs are the same, so there need to be compromises.

Agreed, but don't lose the flip side.  I am not sure if a simple 
database interface will make it into the standard this time around.  In 
Ada 9X it would have been a serious mistake to include one.  This time 
it may be a mistake not to include one, but we could still goof by 
choosing the wrong one.  This is one of the reasons I am participating 
in this discussion.  I would much rather get the key issues for database 
interfaces right for Ada 200X, and allow multiple implementations of the 
actual database interfaces.

-- 
                                                     Robert I. Eachus

"Quality is the Buddha. Quality is scientific reality. Quality is the 
goal of Art. It remains to work these concepts into a practical, 
down-to-earth context, and for this there is nothing more practical or 
down-to-earth than what I have been talking about all along...the repair 
of an old motorcycle."  -- from Zen and the Art of Motorcycle 
Maintenance by Robert Pirsig