Stream: compiler development

Topic: WITS 2024 Proposals


view this post on Zulip Richard Feldman (Nov 09 2023 at 03:33):

I just found out about this, and the deadline to submit is Friday: https://discourse.haskell.org/t/cfp-workshop-on-the-implementation-of-type-systems-wits-24/7873

if anyone wants to submit something I'd be happy to help you get an abstract in at the last minute!

cc @Ayaz Hafiz @Folkert de Vries

view this post on Zulip Richard Feldman (Nov 09 2023 at 03:34):

for example, I suspect if someone were to give a talk about our experiences with lambda sets (which seems extremely relevant to what they're looking for), we'd get a massive boost in relevant interest from people who might want to help out with it

view this post on Zulip Richard Feldman (Nov 09 2023 at 03:35):

The goal of this workshop is to bring together the implementors of a variety of languages with advanced type systems. The main focus is on the practical issues that come up in the implementation of these systems, rather than the theoretical frameworks that underlie them. In particular, we want to encourage exchanging ideas between the communities around specific systems that would otherwise be accessible to only a very select group.

view this post on Zulip Ayaz Hafiz (Nov 09 2023 at 03:41):

yeah we should submit something

view this post on Zulip Ayaz Hafiz (Nov 09 2023 at 03:41):

maybe one talk abstract, one round table proposal

view this post on Zulip Ayaz Hafiz (Nov 09 2023 at 03:43):

my top 3 candidates are lambda sets, emulating subtyping in a unification based type system with a monomorphizing compiler, and the implementation of (algebraic) effects via continuation passing style without GADTs

view this post on Zulip Ayaz Hafiz (Nov 09 2023 at 03:44):

compact and useful error messages in a structurally typed system is also a good one

view this post on Zulip Richard Feldman (Nov 09 2023 at 04:02):

#3 would be a bold submission considering we haven't actually done it yet :laughing:

view this post on Zulip Richard Feldman (Nov 09 2023 at 04:04):

the subtyping emulating could be interesting, although it feels like maybe there's less to say there than lambda sets?

view this post on Zulip Richard Feldman (Nov 09 2023 at 04:04):

I guess if you got into the flow-based inference idea that could be different

view this post on Zulip Ayaz Hafiz (Nov 09 2023 at 04:29):

we could talk about it tho

view this post on Zulip Ayaz Hafiz (Nov 09 2023 at 04:29):

and yeah

view this post on Zulip Ayaz Hafiz (Nov 09 2023 at 04:31):

Richard, what's your connection to the organizers?

view this post on Zulip Ayaz Hafiz (Nov 09 2023 at 04:31):

Like can we hand them 4 proposals or should i wittle it down

view this post on Zulip Richard Feldman (Nov 09 2023 at 04:32):

I know Richard Eisenberg, I'll send him an email

view this post on Zulip Richard Feldman (Nov 09 2023 at 04:33):

but I mean...do we have time to make 4 when the last day to submit is tomorrow? :sweat_smile:

view this post on Zulip Ayaz Hafiz (Nov 09 2023 at 04:47):

Oh it's tomorrow

view this post on Zulip Ayaz Hafiz (Nov 09 2023 at 04:47):

i totally thought 2 days

view this post on Zulip Ayaz Hafiz (Nov 09 2023 at 04:47):

wait no last day is Friday right?

view this post on Zulip Ayaz Hafiz (Nov 09 2023 at 04:47):

I mean what Im suggesting is I'll write two and repurpose both as a talk and a roundtable proposal

view this post on Zulip Ayaz Hafiz (Nov 09 2023 at 04:48):

if someone can help write a couple that would also be great

view this post on Zulip Notification Bot (Nov 09 2023 at 05:19):

21 messages were moved here from #compiler development > Casual Conversation by Ayaz Hafiz.

view this post on Zulip Richard Feldman (Nov 09 2023 at 16:41):

yeah go for it!

view this post on Zulip Richard Feldman (Nov 09 2023 at 16:41):

I don't know how to do the LaTeX formatting but I can help write something

view this post on Zulip Richard Feldman (Nov 09 2023 at 16:41):

Richard Eisenberg said multiple submissions are fine

view this post on Zulip Richard Feldman (Nov 09 2023 at 16:42):

lmk what you'd like me to write and I can take it on!

view this post on Zulip Ayaz Hafiz (Nov 10 2023 at 02:45):

ACM-SIGPLAN-Proceedings-Template.pdf
First one

view this post on Zulip Ayaz Hafiz (Nov 10 2023 at 02:46):

please give me feedback :D

view this post on Zulip Ayaz Hafiz (Nov 10 2023 at 03:18):

ACM-SIGPLAN-Proceedings-Template-2..pdf
number 2

view this post on Zulip Ayaz Hafiz (Nov 10 2023 at 03:43):

ACM-SIGPLAN-Proceedings-Template-3..pdf
number 3

view this post on Zulip Ayaz Hafiz (Nov 10 2023 at 03:43):

fyi @Richard Feldman

view this post on Zulip Ayaz Hafiz (Nov 10 2023 at 03:43):

Also @Folkert de Vries

view this post on Zulip Richard Feldman (Nov 10 2023 at 03:46):

ooo, reading now!

view this post on Zulip Richard Feldman (Nov 10 2023 at 03:49):

nit: in the first one it says "type classes" in one place and "typeclasses" later
bigger point: should that one at least mention challenges we've run into? I think if I were signing up for that talk and it turned out there were unsolved implementation challenges, I'd be surprised based on the abstract :big_smile:

view this post on Zulip Richard Feldman (Nov 10 2023 at 03:51):

also before submitting, there's some generic boilerplate under the "ACM Reference Format" section (e.g. "Conference acronym
’XX" should presumably be "WITS '24")

view this post on Zulip Richard Feldman (Nov 10 2023 at 03:52):

for the second one, I think it might make the abstract more appealing if you mention that we have a before-and-after case study, where we went from HM without polarity to adding it, and saw a dramatic drop in beginner questions about how tag union type variables worked

view this post on Zulip Ayaz Hafiz (Nov 10 2023 at 03:54):

should that one at least mention challenges we've run into?

I don't really know where to put it/how to phrase it

view this post on Zulip Ayaz Hafiz (Nov 10 2023 at 03:55):

for the second one, I think it might make the abstract more appealing if you mention that we have a before-and-after case study, where we went from HM without polarity to adding it, and saw a dramatic drop in beginner questions about how tag union type variables worked

Do you have a suggestion for what the copy should read with that?

view this post on Zulip Richard Feldman (Nov 10 2023 at 03:55):

for the third one, I'd personally mention error union accumulation because that's the biggest selling point of polymorphic variants in effects to me, although maybe an academic audience doesn't care about that because it can already be done in OCaml? :big_smile:

view this post on Zulip Richard Feldman (Nov 10 2023 at 03:57):

Ayaz Hafiz said:

should that one at least mention challenges we've run into?

I don't really know where to put it/how to phrase it

how about this?

We describe our implementation effort to efficiently infer lambda sets, which can be modeled as anonymous sums, including challenges we have encountered in the process—particularly around recursive lambda sets.

view this post on Zulip Richard Feldman (Nov 10 2023 at 03:59):

Ayaz Hafiz said:

for the second one, I think it might make the abstract more appealing if you mention that we have a before-and-after case study, where we went from HM without polarity to adding it, and saw a dramatic drop in beginner questions about how tag union type variables worked

Do you have a suggestion for what the copy should read with that?

how's this?

We will discuss to what extent row and variant polymorphism can aid these goals, using as a case study the Roc programming language's experiences with beginner questions—before and after modifying the language's unification of polymorphic variants to more closely resemble subtyping.

view this post on Zulip Ayaz Hafiz (Nov 10 2023 at 04:15):

Cool cool

view this post on Zulip Ayaz Hafiz (Nov 10 2023 at 04:15):

Im away for a bit but I'll get this to you by midnight CT

view this post on Zulip Richard Feldman (Nov 10 2023 at 04:16):

personally I think you can go ahead and ship it!

view this post on Zulip Richard Feldman (Nov 10 2023 at 04:16):

these look awesome, really great work on them! :smiley:

view this post on Zulip Ayaz Hafiz (Nov 10 2023 at 04:16):

do you want to shoot it over to eisenberg? or did he say to submit it on the platform they have

view this post on Zulip Richard Feldman (Nov 10 2023 at 04:19):

yeah just submit on the platform they have

view this post on Zulip Richard Feldman (Nov 10 2023 at 04:19):

lmk when it's in and I'll double check with him that they received it :big_smile:

view this post on Zulip Richard Feldman (Nov 30 2023 at 01:33):

I got an email with a question:

We're in the midst of building the program for WITS. Your submissions are all exciting. But three talks in one day from one speaker isn't ideal. Each submission has three authors; is it the intent that each author would present a different talk?

personally, I definitely won't be able to make it to the conference - @Ayaz Hafiz I'm assuming you'd be up for presenting? @Folkert de Vries would you be interested and able to attend and present?

view this post on Zulip Ayaz Hafiz (Nov 30 2023 at 01:36):

I would be up, but not 100% sure that I will actually be able to make it. For me, just one is more than enough, I definitely cannot prepare more than that.

view this post on Zulip Richard Feldman (Nov 30 2023 at 01:40):

gotcha - do you know when you'd be able to confirm whether you'd be able to make it? I assume it would not be great for the organizers if they accepted a talk and then we didn't have a presenter for it :sweat_smile:

view this post on Zulip Richard Feldman (Nov 30 2023 at 01:41):

also I suspect that a big part of the value of the conference would be actually being there in person, meeting people, sharing knowledge, making connections for future questions we might have etc.

view this post on Zulip Ayaz Hafiz (Nov 30 2023 at 01:41):

Well, the null hypothesis is that I will attend

view this post on Zulip Ayaz Hafiz (Nov 30 2023 at 01:41):

But I'm not going to commit to it until I get accepted, right

view this post on Zulip Ayaz Hafiz (Nov 30 2023 at 01:41):

I'm not going to attend if I am not speaking is what I mean

view this post on Zulip Richard Feldman (Nov 30 2023 at 01:44):

oh totally haha

view this post on Zulip Richard Feldman (Nov 30 2023 at 01:45):

oh yeah and @Folkert de Vries when you see this - are you up for presenting 0, 1, or 2 talks assuming they're accepted? :big_smile:

view this post on Zulip Folkert de Vries (Nov 30 2023 at 11:22):

realistically, no. The timing does not really work out for me here

view this post on Zulip Richard Feldman (Nov 30 2023 at 15:17):

ok cool, so I'll tell them that we have 1 person who can present if one is accepted. @Ayaz Hafiz do you have a preference for which one of the 3 you'd most like to talk about? I can pass that along :big_smile:

view this post on Zulip Ayaz Hafiz (Nov 30 2023 at 15:25):

Probably "Type-directed defunctionalization in the presence of type classes and monomorphization"

view this post on Zulip Richard Feldman (Dec 04 2023 at 11:56):

all 3 proposals were accepted! :tada:

view this post on Zulip Richard Feldman (Dec 04 2023 at 11:56):

(even though they can't all be presented haha)


Last updated: Jul 06 2025 at 12:14 UTC