Re: Ada Core Technologies and Ada95 Standards

[john@assen.demon.co.uk (John McCabe)]

dewar@cs.nyu.edu (Robert Dewar) wrote:

>John McCabe says

>"This paragraph confuses me.

>I believe that if the compiler vendors were producing high quality
>products that conformed to the language, then passing the ACVC suite
>should be a cinch.

>What you seem to be saying is that compiler vendors are assigning
>resources to ensure that they pass the ACVC suite, and in doing so are
>compromising the quality of their product."

>This is a very common misconception among people who don't know language
>design or compiler implementation very well. Indeed it is the same
>misconception that has lead people to misinterprete what conformance 
>testing is all about.

>Perhaps I can put it this way. Suppose a vendor has resources to do exacty
>one of the following two tasks:

>1. Rewrite the loop optimizer so that all loops run faster

>2. Rewrite the handling of static expressions so that one very obscure
>test in the ACVC suite which has never shown up in a custoer programer
>and is very unlikely *ever* to show up in customer programs, under the
>condition that this rewriting is extensive and will likely cause
>regressions (in programs other than ACVC tests).

>Which do YOU think would contribute more to quality for most users?

I cannot speak for most users, only for myself.

You seem to be defending a "make it fast, THEN make it work"
philosophy here (which I completely disagree with) and confusing
quality with run-time performance.

I have seen the effect of this kind of philosophy - my current
compiler supports the ATAC co-processor, yet has had trouble compiling
some basic Ada constructs.

I would choose 2. over 1. as I believe a _quality_ compiler is one
which compiles the language completely, not one which does some parts
of the language quickly.

What you don't appear to appreciate in your statements here is the
cost involved in tracing the cause of a fault (i.e. a compiler bug) at
the end user level, and then trying to find a workaround solution.
Many bugs are extremely obscure and hard to track down - performance
enhancement at an end user level, while non-trivial in many cases, are
often fairly straightforward in my experience.

>That's the potential trouble with mandated testing, it elevates very
>unimportant problems to maximum priority without any regard to the
>importance or impact of these problems. 

Importance, in this case, is relative - I may be the only customer who
uses the technique covered by the "obscure" test in which case it is
_very_important to me. The point of the ACVC suite etc is to attempt
to ensure that compilers conform to the language standard, is it not.
As there appear to be no "priority levels" defined for language
features, then each is equally important.


Best Regards
John McCabe <john@assen.demon.co.uk>