Subject: Re: [sv-ec] fork..join_none/join_any and automatic variables
From: Steven Sharp (sharp@cadence.com)
Date: Thu May 22 2003 - 16:06:30 PDT
Randy,
The scope of my comments on this was intended to be limited. I really
didn't want to get into a full design discussion about how this issue
should be resolved in SV.
Some suggestions had been made about the behavior of fork-join_none/join_any
with automatic variables, extrapolating from an incorrect concept of the
behavior of fork-join and automatic variables in IEEE 1364-2001. My main
intent was to correct this misunderstanding of the IEEE standard. I wanted
to prevent it from spreading, and to avoid an extrapolation of SV behavior
that would be inconsistent with the 1364 standard.
I also suggested some alternative extrapolations for SV from the correct
1364 behavior. Knowing the rationale behind the specification makes it
easier to extrapolate it to new situations. Also, most participants might
not have been aware that there was already a general rule in 1364-2001
covering subprocesses surviving longer than the surrounding automatic scopes.
In the absence of any resolution in the SV specification, it is the only
existing rule that applies.
It isn't a clear resolution or necessarily the best one. What you have
raised is a serious problem in the specification, and one for which the
best solution is not obvious.
Any further issues involving the 1364-2001 standard should probably be
addressed to the ETF, and not discussed here unless they are relevant to
the SV specification.
Steven Sharp
sharp@cadence.com
This archive was generated by hypermail 2b28 : Thu May 22 2003 - 16:13:21 PDT