RE: [sv-ec] Re: [sv-bc] explicit package exports

Shalom,

 

I believe your assessment is incorrect. I remember discussing the issue
of re-exporting from a package 

as early as 2003 when the "separate compilation" WG began the work that
concluded with the introduction 

of packages into the language.

Allowing exports of imported items were definitely discussed. My
recollection is that at the time, there were 

no objections to the feature, but we would have to work out the details
at a later time. I believe that this topic 

fell between the cracks due to schedule pressures, and left various
committee members with different 

impressions. I always assumed that this was a useful feature that users
would appreciate, and believed other 

members felt that way too. I'm somewhat puzzled that this issue seems so
controversial now.

 

The minutes of at least one of the "separate compilation" meetings shows
this issue was explicitly raised:

 

http://www.eda.org/sv-bc/sep_comp/svbc_03_09_23.txt

 

2) Within a package, are nested references to imported packages exported

        from the package into an importing region?  We'll discuss this
next week.

 

The above minutes are consistent with my recollection that this issue
was discussed, was never explicitly voted 

down, and apparently never worked on. 

 

Since some of us were left with the impression that re-exporting
imported identifiers was an acceptable feature,

the fact that the LRM does not have a contradictory example simply
served to reinforce that conception (or, 

should I say misconception?). I assume that the converse is true for
people who had the opposite impression.

 

I hope this helps clarify the discussion regarding original intent.

 

            Arturo

 

-----Original Message-----
From: owner-sv-bc@eda.org [mailto:owner-sv-bc@eda.org] On Behalf Of
Bresticker, Shalom
Sent: Saturday, September 16, 2006 1:03 PM
To: Brad Pierce; sv-bc@eda-stds.org; SV_EC List; sv-ac@verilog.org
Subject: RE: [sv-ec] Re: [sv-bc] explicit package exports

 

That's my point. There are no LRM examples showing the case one way or

the other because it was never considered, which means that the

functionality was not intended. Adding functionality requires a positive

intention. Lack of an explicit negative intention is not enough. The LRM

does not say that writing "A=1;" does not set off a global thermonuclear

explosion, either, but I don't think that was intended...

 

Shalom

 

> If there were really any LRM examples that "clearly imply that import

> chaining was not intended", then testing with the LRM examples would

> have revealed a mismatch between the intent and what seems obvious

under

> the file system analogy.