RE: [sv-ec] Updated Mantis 1447

Mike,
 
I would not be in favor of removing that DPI-oriented restriction
until implementation teams could study the issues and make
sure there were no severe difficulties.
 
The use of unsized output array arguments is so that a DPI call
can be made with a variety of differently sized fixed-size arrays
as outputs.
 
Regards,
Doug

________________________________

From: owner-sv-ec@server.eda.org on behalf of Michael Burns
Sent: Wed 11/28/2007 5:40 PM
To: Jonathan Bromley
Cc: sv-ec@server.eda.org
Subject: Re: [sv-ec] Updated Mantis 1447



Hi Jonathan,

Thanks for putting that doc together for me. I've incorporated the changes and will be uploading a new version of 1447 shortly. Of course, I made the changes modulo my new text, so the changes aren't exactly what you wrote, but the idea is the same - mostly adding "and queues" in a bunch of places.

I have one question (for everyone). At the end of 7.7 "Arrays as arguments..." it speaks of passing dynamic array (and queue) actuals to DPI formals. In particular, the following sentence appears in this proposal (note that I was not the original author, so I don't know why it's there):


A dynamic array or queue cannot shall not be passed as an actual argument if the DPI formal argument has is an unsized dimensions and an output direction mode.

Is this right? Do we want this restriction? What is the reason for it? Is it possible that there is a mistake here, and that we really only want to have this restriction for unsized _packed_ dimensions of output formals? If we cannot use dynamic arrays and queues here, what is the purpose of having DPI unsized output array arguments?

Thanks,
Mike


Jonathan Bromley wrote: 

	Mike,
	
	  

			Would you be open to an adjustment of 1447 that would clarify
			assignment-compatibility amongst all kinds of unpacked array
			except associative (i.e. dynamic, queue and fixed-size)?  
			If so, I could restrict the scope of 1702 to affect only clause 10.
			      

	  

		Sure - let me know what changes you'd like to make and I can 
		incorporate them into 1447. Assuming there are no land mines,
		of course :) 
		I need to make another pass anyways to incorporate
		Shalom's feedback.
		    

	I've split my 1702 proposal so as to decouple it from 
	your 1447.  I'd tried to make a few changes to clause 7
	to do much the same as 1447 does, but I was much less
	thorough about it than you have been.  However, your
	1447 proposal doesn't say much about queues, and I think 
	there is a consensus that your treatment of dynamic
	dimensions of an array is pretty much applicable to 
	queues also (except, of course, that new[] doesn't
	apply to queues).
	
	Here's a document (both PDF and MS Word) containing
	the 1447-ish changes that I've now removed from 1702.
	My guess is that you'll decide you have dealt with 
	almost all of it already, except that you'll need 
	to say "and queues" in a few places....
	
	Please shout if there's anything you need me to do 
	about this.
	
	Thanks and regards
	  


-- 
This message has been scanned for viruses and 
dangerous content by MailScanner <http://www.mailscanner.info/> , and is 
believed to be clean. 

-- 
This message has been scanned for viruses and
dangerous content by MailScanner, and is
believed to be clean.