RE: [sv-cc] Draft 3 of the LRM is now available

Subject: RE: [sv-cc] Draft 3 of the LRM is now available
From: Warmke, Doug (doug_warmke@mentorg.com)
Date: Fri Jan 09 2004 - 16:30:49 PST

Next message: David W. Smith: "RE: [sv-cc] Draft 3 of the LRM is now available"
Previous message: David W. Smith: "[sv-cc] Draft 3 of the LRM is now available"
Maybe in reply to: David W. Smith: "[sv-cc] Draft 3 of the LRM is now available"
Reply: David W. Smith: "RE: [sv-cc] Draft 3 of the LRM is now available"
Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]

David, Stu,

Thanks for the monumental amount of work you put into the LRM.
Here are some feedback comments on draft3.

Section 27:

1. The comment in the Editor's note in section 27.3 is fine. The note
can be removed.
2. Section 27.4.1 was changed "functions" is now "function calls".
   This change is not correct and should be backed out.
   The title should be "Required properties of imported tasks and
functions - semantic constraints"
3. Regarding the Editor's note in section 27.4.1.3, David is correct.
   There are three types of imported task or function:
    a. pure
    b. "normal" (aka non-context)
    c. context
   In order to clarify Stu's concerns, I would suggest placing the
following paragraph
   exactly where the editor's note is:
      "If neither the pure nor the context attribute is used on an
imported task or function,
       the task or function may not access SystemVerilog data objects,
however it can
       perform side-effects such as writing to a file or manipulating a
global variable."
   Hopefully this clarifies that there is a "third way" for these
functions.
   I would be fine leaving this out, too, if Stu changes his mind or
other people object.
4. Section 27.4.3 should have its title changed to "Context tasks and
functions"
5. Regarding the first Editor's note in section 27.4.3, I agree with
Stu.
   Please remove the mentioned text.
6. Regarding the second Editor's note, same thing as above. Please
change "tasks or functions"
   to just "functions".
7. I would suggest dropping the first Editor's note in section 27.4.4
for now.
   One point is that chapter 27 only deals with the SystemVerilog side
of the interface,
   and void functions are a reality in SystemVerilog. If we wanted to
add support for
   a foreign language other than C/C++, we could handle the details of
correspondence
   with SV void functions in the appropriate Annex.
8. Good idea with the second Edtior's note in 27.4.4.
   In addition, I would recommend adding footnotes 9,10 into that inset
box.
   Those are really significant (partly because they outlaw the
tf_ref_declaration
   that was introduced with the tf_proto_formal production).
9. The Editor's note in section 27.8 is fine. It can be removed.
10. Item 4) in the protocol in 27.8 should have a minor change made to
it:
    "..., it is illegal to make any further calls..." should be changed
to
    "..., it is illegal for the current function invocation to make any
further calls..."

Annex E:
1. The editor's note in E.8.3 is correct. The second addition of "tasks
or" should be removed.

Thanks again for all the work.

Regards,
Doug