Subject: [sv-cc] Complete meeting minutes of 01/28/03
From: Swapnajit Mittra (mittra@juno.com)
Date: Tue Jan 28 2003 - 19:09:58 PST
Complete meeting minutes for 01/28/03.
(Thanks Michael and John A!). Please
send your correction to the team or bring
them up during the next meeting.
- Swapnajit.
Attendances:
Francoise Martinole
John Amouroux
Doug Warmke
Michael Rohleder
Kevin Cameron
Bassam Tabbara
Swapnajit Mittra
Joao Geada
Ghassan Khoory
Andrzej Litwiniuk
Joe Daniels
Stuart Swan
1. Issue 1.9 (C/C++ code inclusion):
Michael revisited the changes he made to the inclusion document as a result of the last F2F meeting.
He did split the document into two pieces; actually three: an introduction, a first part that defines the inclusion of object code,
and a second part that defines the inclusion of source code. The
rationale for the introduction is that it contains common statements
and might be useful when we add a part about PLI/VPI inclusion at a
later point in time.
Changes to the introduction:
. added a switch '-sv_root' to mirror the SV_ROOT environment variable as
requested by Joao.
. added a general statement on how to handle environment variables
Changes to the first part
. There were only minor modifications to this part to make it consistent with
other modifications.
. Also an additional example has been added
Changes to the second part
. The second part has been completely restructured. Switches have been added to
mirror the functionality
provided by environment variables (as requested in the last meeting).
. The assumed compilation command structure is now identified to explain which
parts are replaced.
. Additional switches are now provided to permit overriding the compiler options for defining include file directories, input file
and output file. Although all compilers Michael is aware of use the same switches for defining these items, he thinks it is safer to permit overriding these items.
. One more example has been added to this part
[until then there were no comments from other team members]
Michael then started to discuss the comments he has received until now from Francoise and John.
. Multiple occurances of files: (Francoise)
Michael referred to his email response and explained why he wants to avoid
linker errors on multiple
occurances of the same file. Upon request from Michael Francoise stated that
she understands his
rationale, but is unsure whether such checks are needed at all. Joao asked
why it would not be possible
to detect even the duplicates in case of links, as Michael described in his
response, by checking the inodes.
Kevin had concerns that this will not work in Windows. Michael and Joao
stated that links/references in
Windows are significantly different from links in UNIX.
The agreement was to amend the proposal to cover cases where it is possible
to identify duplicates.
. One solutions vs. three levels of Flexibility (Francoise)
There was a very lifely discussion about the proposed amount of flexibility
in the proposal.
- Francoise is concerned about the amount of flexibility and would like to
have less
- Kevin is unsure whether we should not cover more important things.
- Joao, Doug & John are concerned about the side effects possible due to
environment variables.
Since there was no strong opinion about the need for environment variables,
the agreement was to eliminate
the usage of environment variables whenever possible.
. Default value for sv_root (John)
Michael agreed to John's proposal. This is clearly an omission of the
proposal.
. Usage of file identifiers as described
John mentioned that the 'file' command does not show the file type; it has
been clarified that this requires
some update in the descripton database of the 'file' command. Although some
members question the
use of these identifiers, there was no strong opinion against it.
. Syntax of the SV_LIBRARIES environment variable
Michael stated that this issue is no longer of interest when environment
variables are gone.
Three more issues brought up during the discussion where:
. Do we really need to support multiple occurances of -sv_root? (Doug)
Michael answered yes. There are often multiple users/projects sharing code
and those are having
highly distributed databases.
. Currently we are only permit archives and shared libraries, why not the
inclusion of .o files? (Andrzej)
- Michael: this is intentionally, since linking of archives removes
unreferences functions
- John (?) inclusion of .o files would also help in the most simple case
- Francoise: only the loading of shared libraries is clearly defined
currently by C/C++
- Doug: currently only loading of shared libraries is possible in NC and MTI,
only VCS
can load archives (and might be able to do so with .o). Do we add
additional complexity
for user convinience
It has been decided to put this discussion to the reflector.
. How is it possible to associate SV source code related to the compiled
objects and how to locate this (Andrzej)
There was some discussion about how this can/should be handled. John
requested some use cases for this.
It has also been decided to put this discussion on the reflector.
2. Coverage API Changes:
Joao presents his changes - his doc has all the change bars set.
Basically, he has fixed up the doc according to the previous input from the group.
Andrzej: How does the database file relate to checkpoint save/restart?
Joao: These do not interact with each other. But a simulation doing a
checkpoint save should also save coverage data. But the coverage data saved
in the checkpoint file would not be visible to the user.
Michael: coverage_get_max omission
Joao: forgot to add Michael's suggestions - will do it in next rev of the doc.
Poll for coverage acceptance slated for Feb. 6, depending on the feedback we see on the reflector.
Joao needs to do the updates and send them to Joe as soon as possible.
-- Swapnajit Mittra Project VeriPage ::: http://www.angelfire.com/ca/verilog________________________________________________________________ Sign Up for Juno Platinum Internet Access Today Only $9.95 per month! Visit www.juno.com
This archive was generated by hypermail 2b28 : Tue Jan 28 2003 - 19:11:34 PST