Re: Ada Core Technologies and Ada95 Standards

[dewar@cs.nyu.edu (Robert Dewar)]

John McCabe said

  "I assume by point (c) that you mean compilers must not contain
  improper extensions. That is an interesting point to me since in
  another thread somewhere we discussed TLD's implementation of package
  Machine_Code. Your response to my description of their implementation
  was that this was precisely an improper extension. As this
  implementation ahs been around for donkeys years, how did it get
  through the validation process with an improper extension as obvious
  as that?"

First of all, as is clearly stated in the validation procedures document
(I stronly recommend reading these procedures if you want to make informed
comments on the validation procedures), the approach for eliminating
extensions is the DOC (declaration of conformance). Obviously no black
box testing can detect extensions in a systematic way, and even examining
the code is an impractical and ineffective way of making extensions. The
DOC actually declares that you have no deliberate extensions.

So TLD does not think that this unusual use of Machine_Code is an extension.
That's arguable on either side. I would consider it an extension, but I am
not in authority here. I would assume the argument on TLD's side would go
something like this: Package Machine_Code as described in the RM is
optional, we don't implement it. We are allowed to add packages (adding
packages is not an extension). We have added a package which happens to
have the same name as the optional package in the RM we do not implement.
Where is this prohibited? We do not think it is prohibited, and since
the functionality, if not the form, of our package matches the intent of
the RM defined package, it seemed a reasonable name. I guess that you
would have to agree with this argument, although it would probably take
a WG9 ruling to be sure. I still prefer not to confuse the use of the
package name here, but thinking about it more, I agree that this is not
an extension in the formal sense. Informally it feels like an extension
to me, but I have to admit that I cannot prove it.
  
  "I fear you may be biased towards this type of bug reporting by being
  directly involved with the Ada language itself. How many [other]
  compiler vendors are this conscientious?"

Not quite sure what this means. All Ada implementors are involved with
the Ada language. I am not the only vendor with people on various
committees (I do not say representatives here, because none of us
represent our companies in this context). Many suggestions for tests
in the past have come from vendors. I can't speak for how concientious
any vendor may or may not have been in this regard.

    >P.S. I find it a bit amazing that John McCabe is so unaware of the
    >validation status of the compiler he is using. One important piece
    >of advice for any user of validatd Ada compilers is to obtain the
    >VSR (validation status report), and read it carefully. VSR's are
    >public documents, available from the AVO, so even if your vendor
    >does not supply a copy (they should), you can obtain one. John,
    >along with a lot of other data, the VSR lists the expiration date,
    >or points to the documents that define the expiration date.
  
  "I don't find it amazing at all! As I have probably mentioned in this
  and other threads, I am mandated to use this version of the compiler
  by my customer's customer's customer so I don't really give a s**t
  about its validation status.

Well you are free to take the "I don't really give a s**t" attitude to
anything you like, but the fact of the matter is that the validation
status report (VSR) has valuable information about any compiler, and is
often provides information that is valuable in using a compiler. Also
I wish to remind you that it was you who raised the issue of the validation
status of the compiler you are using, not me :-)

P.S. do you like that indentation style better?