Today's meeting and late notes for Feb 4 meeting

[Hans Boehm]

We will be meeting today, Mar 4, at 12:00 PST

meet.google.com/sbj-cvgh-vnd

or

+1 208-925-0196‬ PIN: ‪255 542‬#

The meeting is likely to be brief unless there is progress to report.

Feb. 4 notes:

Attendees:

Hans Boehm (HB), Michael Scott (MSC), Mike Spear(MSP), Victor Luchangco(VL)

What's a quorum? Pursue a paper in either case.

No major progress on proposal.

MSP: Just paper? Or implementation?

VL: Paper first

MSP. What's actually in the proposal

MSC: Fairly comprehensive.

Based on constexpr

Constexpr ==> in same translation unit?

HB: Dynamically invoke only constexpr functions.

VL: At least close

MSC : Parameters can be dynamic

Constexpr can throw.

Go with UB if body is not in the right subset.

Make it as easy to implement as possible.

Previously agreed:

No explicit restriction on Transaction size.

MSP: Current implementation handles basically everything but keyword.

MS:. Advantage of syntax: Easier to optimize read-only transactions.

Less of an issue for lock elisiion.

Deletion from current spec.

VL: Need separate conceptual paper.

Memory model? Same as current proposal (N4514) Non-conflicting transactions are not ordered. Empty transactions can be elided. Deletion from current document should give the right answer.

Transaction-safe keyword goes away.

Next meeting: March 4