P2066 Suggested draft TS for C++ Extensions for Transaction Memory Light

[wg21bot]

P2066R0 Suggested draft TS for C++ Extensions for Transaction Memory Light (Jens Maurer, Michael L. Scott)

[jfbastien]

Let's have SG1 take a look first, then forward to EWG.

[ogiroux]

SG1 in Prague

We will see this again in Varna

[wg21bot]

P2066R1 Suggested draft TS for C++ Extensions for Transaction Memory Light (Jens Maurer, Hans Boehm, Victor Luchangco, Michael L. Scott, Michael Spear, and Michael Wong)

[wg21bot]

P2066R2 Suggested draft TS for C++ Extensions for Transaction Memory Light (Jens Maurer, Hans Boehm, Victor Luchangco, Michael L. Scott, Michael Spear, and Michael Wong)

[ogiroux]

SG1 approves the design and wording of P2066R2 targeting the next
revision of the TM TS (which is not expected to be reaffirmed in its
earlier form).
No objection to unanimous consent. (15 people attending)

[jensmaurer]

This contains no library changes, but only core language changes, and thus should go to EWG next.

[jfbastien]

This was discussed in EWG telecon today.

Use a/many (potentially contextual) keyword to denote atomic blocks, rather than making it a library feature. For example atomic do { ... }.

SF	F	N	A	SA
5	8	1	0	0

The proposed approach to TM generally seems acceptable, and would fit in a TM TS v2.

SF	F	N	A	SA
3	12	1	0	0

We'll see it again after discussing the particular keyword approach. LEWG should also see it to discuss library support guarantees, if any.

[wg21bot]

P2066R3 Suggested draft TS for C++ Extensions for Transaction Memory Light (Jens Maurer, Hans Boehm, Victor Luchangco, Michael L. Scott, Michael Spear, and Michael Wong)

[wg21bot]

P2066R4 Suggested draft TS for C++ Extensions for Minimal Transactional Memory (Jens Maurer, Hans Boehm, Victor Luchangco, Michael L. Scott, Michael Spear, and Michael Wong)