Subject: Oct 20 minutes
From: Hans Boehm (boehm_at_[hidden])
Date: 2020-10-20 20:16:45
Attendees: Hans Boehm, Victor Luchangco, Jens Maurer (at the beginning),
Michael Scott, Mike Spear, and Michael Wong
Initially discussed p2066r4 and vector::size(), and the question of whether
it is and should be included in the set of guaranteed transaction-safe
We agreed that the current definition of "accessors" is not clear enough.
At this point, there also seemed to be close to consensus that we should
try to initially keep things as simple as possible, and avoid anything that
might look scary to implementors.
We then tried to switch back to the discussion of including new.
Mike Spear argued that adding new and supporting non-escaping exceptions
doesn't really complicate simple implementations. Both single-global-lock
and HTM implementations still work.
Mike Spear: What can you actually do with the p2066 proposal?
Michael Scott: implement lock-free data structures like linked lists that
read without a lock, and then transactionally update it at the right place.
Mike Spear: No, because the initial reads can't be on atomic data.
Can't really use these transactions as an NCAS replacement. All read
accesses preceding NCAS would also have to be transactional, which is too
expensive for something like Intel HTM. Transactions cost as much as locks,
so surrounding every read by one is not practical.
Victor L.: Transactional reads could work for tree traversal followed by
update, which is probably more common.
Clarification that this avoids contention in an HTM context, but still has
latency issues even in the uncontended case.
Discussion of double-checked locking, with a similar conclusion. It's not
practical to put the initial read in a transaction.
We unfortunately concluded that we either need more of a usability argument
for what we have, or probably a revision of both the design and wording
documents to allow new + caught exceptions. If we had the latter, most of
the STL data structures could become usable in transactions. Given that
roll-your-own data structures don't seem terribly viable, that would give
us a much better usability argument.
Closed with a strong recommendation to continue this discussion online, and
not wait until the next meeting.
Nov. 17, 8:00 PST, 11:00 EST
Join with Google Meet
Join by phone
âª(US) +1 208-925-0196â¬ PIN: âª255 542â¬#
SG5 list run by firstname.lastname@example.org
Older Archives on Google Groups