Re: DCL improvements

_at_SendSpamHere.ORG
Date: 10/07/03 

Date: Tue, 07 Oct 2003 15:51:19 GMT

In article <sd80wS3zf2Ey@eisner.encompasserve.org>, briggs@encompasserve.org writes:
>In article <6OCT03.21142891@feda01.fed.ornl.gov>, Dave Greenwood <greenwoodde@ornl.gov> writes:
>> In a previous article, Guy Peleg <guy.peleg@hp.com> wrote:
>>> Hi Hans,
>>>
>>> I'm currently working on WHILE - expected release is V8.2.
>>>
>>> As for F$GETDVI("AVL") I agree that FALSE should be returned. Could you please
>>> log
>>> a call with your local support center and have it escalated to engineering? I
>>> will provide you a fix to your
>>> current version and will make sure to include it in V8.2.
>>
>> I'll have to agree with the previously noted concern about breaking
>> existing code with this change. I'll also note that if you made the
>> change for AVL I'd expect you to make the same change for *all* the
>> getdvi item codes.
>
>Which means determining a default value to return for all those codes
>that is reasonable for a non-existent device. Not all _that_ hard to
>do, I suppose. "False", 0 and "" should serve for most everything.
>
>We've had this argument before.
>
>The purist point of view is that asking for any attribute other than
>"EXISTS" of a non-existent device is a nonsensical thing to do. The
>lexical function should mirror the operation of the underlying system
>service. And the underlying system service fails with an error code
>of SS$_NOSUCHDEV.

Being the "purist", where will this change be implemented? In the $GETDVI
service code or just in the lexical?

If I have a procedure that checks for device AVaiLability, what shall I do
when I get back "FALSE" and the code tried to make said device AVaiLable?

>The non-purist point of view is "I want to know if the device is available
>for use. And if it doesn't exist then it's surely not available. Why
>make me waste two lexical calls when one would do?

Anybody writing procedures in DCL is most certainly *NOT* concerned with
CPU cycles it takes to get the job done. If efficiency was a design goal
for the task implemented, DCL would be the first item to be scratched off
of my implementation language list. 

-- 
VAXman- A Bored Certified VMS Kernel Mode Hacker    VAXman(at)TMESIS(dot)COM
           
  "Well my son, life is like a beanstalk, isn't it?"