Stream: t-compiler/wg-nll

Topic: weekly meeting 2019.03.13


pnkfelix (Mar 13 2019 at 19:24, on Zulip):

hi @WG-nll

pnkfelix (Mar 13 2019 at 19:25, on Zulip):

I forgot to do all of: 1. send out a reminder that due to USA daylight savings, our scheduled meeting time may or may not have shifted

pnkfelix (Mar 13 2019 at 19:25, on Zulip):

2. send out a reminder saying that the meeting is starting in N minutes

pnkfelix (Mar 13 2019 at 19:25, on Zulip):

3. do pre-triage

pnkfelix (Mar 13 2019 at 19:26, on Zulip):

I think the vast majority of the people in @WG-nll are actually not based in the USA, which may mean that the USA's out-of-sync DST shift should not affect us anyway...?

pnkfelix (Mar 13 2019 at 19:27, on Zulip):

so lets just find out: How many people would attend if the meeting were held in 3 minutes?

pnkfelix (Mar 13 2019 at 19:27, on Zulip):

(versus how many would attend if the meeting were held in 63 minutes)

pnkfelix (Mar 13 2019 at 19:28, on Zulip):

(and while I await any emoji reactions to those Q's, I'll attempt some pre-triage in a different topic.)

pnkfelix (Mar 13 2019 at 19:33, on Zulip):

we have zero P-high unassigned issues, which is great

pnkfelix (Mar 13 2019 at 19:35, on Zulip):

based on the lack of reactions to the messages above, I'm going to assume (or assert?) that everyone figured the meeting would stick to European DST rather than USA DST, and thus it won't be held for another 55 minutes. This means we may not see @nikomatsakis at the meeting; we'll see.

nikomatsakis (Mar 13 2019 at 19:35, on Zulip):

I may or may not be able to make it but I can catch up later!

nikomatsakis (Mar 13 2019 at 19:36, on Zulip):

:heart: Zulip

pnkfelix (Mar 13 2019 at 19:37, on Zulip):

@nikomatsakis any chance of an update on #54940 in the meantime?

pnkfelix (Mar 13 2019 at 19:39, on Zulip):

also, @Matthew Jasper , I think I have a plausible implemenation of a lint form for PR #58739

nikomatsakis (Mar 13 2019 at 19:39, on Zulip):

Not really. My expectation is that this won't be closed for some time, but looking at it again -- and in light of the 2PB discussion =) -- maybe I should think harder about a way to fix it

pnkfelix (Mar 13 2019 at 19:39, on Zulip):

@Matthew Jasper i assume you won't mind if I push the commits for that onto your branch?

Matthew Jasper (Mar 13 2019 at 19:39, on Zulip):

go ahead

pnkfelix (Mar 13 2019 at 19:40, on Zulip):

Not really. My expectation is that this won't be closed for some time, but looking at it again -- and in light of the 2PB discussion =) -- maybe I should think harder about a way to fix it

fix it in a way that doesn't require lazy-normalization?

pnkfelix (Mar 13 2019 at 19:41, on Zulip):

or think harder about lazy normalization?

pnkfelix (Mar 13 2019 at 19:46, on Zulip):

One thing I did want to bring up here, though maybe its a waste of time to discuss

pnkfelix (Mar 13 2019 at 19:46, on Zulip):

this is regarding PR #58739

pnkfelix (Mar 13 2019 at 19:47, on Zulip):

if you look at one of my recent comments, you'll see i sort of hand-waved about the schedule for when the lint would go to hard error

pnkfelix (Mar 13 2019 at 19:48, on Zulip):

but part of that hand-waving is that I said "we can make the 2pb restriction a hard error on the 2015 edition"

pnkfelix (Mar 13 2019 at 19:48, on Zulip):

I probably already know the likely reactions, but I want to ask the question: What is more important: That the 2015+2018 editions be consistent in their static semantics?

pnkfelix (Mar 13 2019 at 19:49, on Zulip):

Or that we try to limit the amount of code that is written that utilizes the less-restricted 2pb? (And thus make it a hard error immediately for the 2015 edition)

pnkfelix (Mar 13 2019 at 19:50, on Zulip):

I'll admit, I think we even may have talked about this, or something similar, last week and came down on the side of "keep static semantics of 2015 and 2018 consistent"

pnkfelix (Mar 13 2019 at 19:50, on Zulip):

but I just wanted to make sure I brought it up again

centril (Mar 13 2019 at 19:56, on Zulip):

@pnkfelix If we don't have the restriction on 2018 we still need an operational semantics that handles the less restrictive 2pb; in that case it's of zero value to make such a restriction on 2015.

pnkfelix (Mar 13 2019 at 19:56, on Zulip):

oh yeah, we definitely discussed this last week

pnkfelix (Mar 13 2019 at 19:57, on Zulip):

its a funny thing, saying we don't have an operational semantics here

pnkfelix (Mar 13 2019 at 19:59, on Zulip):

well, I go back and forth on the matter

pnkfelix (Mar 13 2019 at 19:59, on Zulip):

my instinct is to say "of course we have an operational semantics, in terms of what the dynamic semantics are for this construct. its just a problem with the elaborated semantics that Ralf is designing."

pnkfelix (Mar 13 2019 at 20:00, on Zulip):

but the point (as in, Centril's point) still stands, once you start thinking about unsafe code

pnkfelix (Mar 13 2019 at 20:01, on Zulip):

(clearly the answer is to disallow casting a reserved unactivated borrow to a *T)

pnkfelix (Mar 13 2019 at 20:01, on Zulip):

/me is all about hacks tonight

Matthew Jasper (Mar 13 2019 at 20:02, on Zulip):

The problem is with the thing that we allow. We already have every other restriction that one could reasonably come up with.

centril (Mar 13 2019 at 20:03, on Zulip):

@pnkfelix I mean... we have a dynamic semantics... :P programs must run... to be clear I meant "a formal one" ;)

pnkfelix (Mar 13 2019 at 20:05, on Zulip):

i know, i know. I just meant that the intuitive meaning of the borrowing constructs here is "obvious."

pnkfelix (Mar 13 2019 at 20:07, on Zulip):

except that its not obvious once you try to encode the borrow restrictions into the dynamic semantics (as to catch bugs that arise when one manipulates unsafe pointers)

Matthew Jasper (Mar 13 2019 at 20:09, on Zulip):

Well, there are ways of doing it, they're just not very good.

pnkfelix (Mar 13 2019 at 20:38, on Zulip):

heh, maybe that was the meeting.

pnkfelix (Mar 13 2019 at 20:39, on Zulip):

/me got distracted with some other things. At this point I think we might as well just have informal office hours with @WG-nll ; I got through everything that I wanted to discuss in the comments above.

Last update: Nov 21 2019 at 14:50UTC