Subject: [sv-cc] Meeting minutes 2/11/03
From: Francoise Martinolle (fm@cadence.com)
Date: Tue Feb 11 2003 - 10:02:56 PST
Attendance:
Michael
Swanaijit
Ghassam
Joao
Kevin
Francoise
Andrzej
Doug
John Amouroux
Bassam
minutes from last week: Michael proposed, Bassam seconds
Organizational:
sv-ac wants to take a look at the Assertion API
Joao required to make a presentation at the beginning of March
Francoise, Kevin
pointer proposal reviewed by the EC? no. They are reviewing the
testbench stuff
Michael:
SV side of C layer? alias name for C function
SV name and C name can be different.
Joao is checking if this requirement was passed to ec: yes it was
For extern and export declarations, we want to be able to alias the SV or C
name.
For externs, it is not a requirement because any C identifier is a legal
Verilog identifier
but it is desired for symmetry
Joao: extern and export already exists in SV.
Doug: if we were going them a proposal, it may make the ec task easier.
Swapnajit: Dave Smith had some questions.
Joao: he wanted:
the requirements and proposal for handles
confirmation that we are okay to use the string class
requirements for extern/export + proposal
Joao is working on a proposal for extern/export.
Kevin: I put a proposal for handles
handles for carrying pointers for systemVerilog
for Direct C, we need to seamlessly be able to call a SV function
Kevin: why not declare the pointer in SV?
Joao: not necessary
Kevin will send out the proposal, discussion to comtinue by email
Swapnajit:
LRM: Joao and Joe are working on the LRM.
Joao: 1 minor review comments round to be sent today to joe before posting
the LRM
to the reflector
Swapnajit:
Friday is the deadline for proof reading of the LRM with the assertion
inclusion.
Andrzej:
Joe sent the revised version of the C layer today. Andrzej needs to review
Swapnajit:
Joe needs to be done by 2/24 and pass it to Stu.
Michael:
any effort to compare and validate SV to C side versus C calling SV?
nothing planned except for reading the final draft.
Continuing the hot pointer discussion...
Doug:
may look like C pointers which looks like they can be dereferenced in SV
opaque handle type may be better
Kevin: I have an issue with syntax drifting away from C.
Joao: the C side is declaring it as a pointer
A pointer in C domain can be a pointer to anything, a pointer to a
function, to a data...
cannot declare any of those in SV
A C pointer is just a void * in SV
Joao: agree with the goals of same prototype in C and SV.
there is no such a header file in C, there is a pre-processor which can be used
on the C header, all this make it difficult to generate the SV prototype.
DirectC is a foreign language interface which uses C.
General language have 3 categories of pointers: pointer to data, to
function, pointer to instances (aggregate of data).
The systemVerilog only knows about 1 : pointer to data
Kevin:
A pointer is an int in the compiler system
Joao: process a C header file is very difficult (typedef,...), type needs
to be known at elaboration time to be able to generate code.
Kevin: a large amount of C will be compatible, would like to have some
compatibility
rather than none
Kevin: the extern declaration can be replaced by the C function prototype.
Kevin: people are going to want to have pointers, the user would have to
make a lot more work to interface to C
Joao: you have the handle type
Discussion to continue by email
Meeting cancelled tomorrow.
Next meeting next week
This archive was generated by hypermail 2b28 : Tue Feb 11 2003 - 10:03:33 PST