Stream: t-compiler/wg-polonius

Topic: meeting 2019.04.11


nikomatsakis (Apr 11 2019 at 20:43, on Zulip):

Sorry I missed the meeting today. I think we should probably find another time.

nikomatsakis (Apr 11 2019 at 20:43, on Zulip):

cc @WG-polonius

nikomatsakis (Apr 11 2019 at 20:44, on Zulip):

But I guess @Albin Stjerna and I did a bit of a sync the other day anyhow

nikomatsakis (Apr 11 2019 at 20:44, on Zulip):

@Keith Yeung -- saw your message. One thing that I think would be helpful is to continue some of the work that @Albin Stjerna was doing on trying to find good benchmarks.

nikomatsakis (Apr 11 2019 at 20:45, on Zulip):

I think as we proceed we'll find other good things to split off

nikomatsakis (Apr 11 2019 at 20:45, on Zulip):

Sorry I missed the meeting today. I think we should probably find another time.

(this time is often a bit chaotic for me)

Keith Yeung (Apr 11 2019 at 20:45, on Zulip):

i presume that by good benchmarks, you mean looking for rust programs that stretches the perf of polonius?

nikomatsakis (Apr 11 2019 at 21:42, on Zulip):

yeah so @Albin Stjerna did some work in this topic doing a bunch of measurements and categorizing them

nikomatsakis (Apr 11 2019 at 21:42, on Zulip):

I think what we want to do is to kind of find some 'good exampes' that fit different definitions

nikomatsakis (Apr 11 2019 at 21:42, on Zulip):

i was going to write up some thoughts on how to do this, e.g. one thought I had is to try and make a spreadsheet with like "how many facts of each kind are there"

nikomatsakis (Apr 11 2019 at 21:43, on Zulip):

I suspect that some benhcmarks have large number of cfg points, other large numbers of subtype relatons, whatever

Keith Yeung (Apr 11 2019 at 23:29, on Zulip):

so, i recall that a couple of the fetch algorithms implemented in servo deals with complicated cases of borrowing

Keith Yeung (Apr 12 2019 at 03:26, on Zulip):

i think this should be it

Keith Yeung (Apr 12 2019 at 03:28, on Zulip):

specifically the assignment to internal_response, based on some other conditions

Albin Stjerna (Apr 12 2019 at 07:22, on Zulip):

Sorry I missed the meeting today. I think we should probably find another time.

yes, with DST it's a bit late for me :(

Albin Stjerna (Apr 12 2019 at 07:24, on Zulip):

@Keith Yeung Also, if you want the things I have made (modifications to rustc-perf, a Jupyter notebook to analyse the results, and my scripts for scraping and compiling crates), just shout!

Last update: Nov 15 2019 at 20:15UTC