[tialaramex]

As I understand it, the larger purpose of this work is as part of efforts to persuade WG14 to pick actual provenance semantics, presumably a PNVI-ae variant. As just a TR, the same as with the original thesis, it doesn't do anything.

Whereas with Rust the point is to actually do something, including in this particular case Aria's Strict Provenance Experiment, which is basically "What if Rust insisted on full blown PNVI?". Obviously the answer is "Well that can't work" but like, how much. How much can't it work? Hence the experiment.

[pjmlp]

I doubt they would get persuaded, as much as they have been into anything related to improving C's safety.

[jcranmer]

From what I understand, the C committee is very interested in adopting pointer provenance semantics, but they also tend to operate at the speed of molasses and this is of course a research-hard problem.

[pjmlp]

Well, in 50 years they have hardly bothered to add any kind of fat pointer as suggested by Dennis Ritchie, Walter Bright and others, struct based vocabulary types for string and arrays, annotations like SAL.

Hence why I hardly believe pointer provenance is going to be any different.

[tialaramex]

Last I looked WG14 intended to knock together a TS for provenance in the C23 timeframe (so, this year). Of course a TS might go nowhere, vendors could ignore it, it might be a dead end, but I don't believe WG14 produced a TS of sane array types, or fat pointers.