Sign in with Twitter

Username:

Occasional OCaml programmer

111 Following   3,410 Followers   2,298 Tweets

Joined Twitter 6/21/09


@fraterrisus @DanielleSucher @NovalisDMT it's more of a code review tool that's a good task manager if all your tasks are writing code...Why doesn't Twitter have an edit button yet? Having to delete and repost on every typo is a drag... @DanielleSucher @NovalisDMT @fraterrisus Yeah, installing complex packages without opam is hard work...This is one of the best aspects of Iron. https://t.co/vIH9vfiuLQ @DanielleSucher @fraterrisus @NovalisDMT @github Also, it only works with hg now, which reduces the audience considerably. @DanielleSucher @fraterrisus @NovalisDMT @github that said, I don't think anyone has tried to set it up externally yet. @DanielleSucher @fraterrisus @NovalisDMT @github That's not a fork, really. Just the public release. It should install via opam...
8/24
2016
@mattmight Since you're in town, reach out if you want to swing by Jane Street and say hello...
8/23
2016
And this shows what the interface of the view function is: https://t.co/lGNOZ1P8Po @czaplic I'm curious what you think of this approach.This is how we avoid having to parameterize our Vdom type. I think this is an improvement over Elm's approach. https://t.co/S6U5n5OW7Q @vsipuli not terribly discoverable, though... @vsipuli https://t.co/jBvFmMHsDQ
8/22
2016
@jmsdnns he's 14 and has been doing racket since he was 9. It's the 9 year old I was looking for advice for. @rudolfsvikmanis 14Of course, the problem of teaching your son something fun about the lambda calculus: who can he tell?Just showed my son the y-combinator. Mind, blown.
8/19
2016
Universities should accelerate innovation. When they give patents to trolls, they do just the opposite. https://t.co/juxLkRU2Dd
Retweeted by Yaron Minsky @triampurum @headinthebox the core observation is that react and elm are functional in nature. That's entirely reasonable. @headinthebox Though you're right that the article is a little muddled at points... @headinthebox So I would tie it more to SAC and FRP. The key goal is optimization of the (purely functional) view function. @headinthebox Also, Rx is the wrong analogy here. React/Elm are more about incremental programming then "reactive" programming. @headinthebox Treating one as functional and one as not just muddies the intellectual waters. @headinthebox Indeed, Elm does effectively the same thing in a language that tracks effects in the type system. @headinthebox React isn't using "Rx", and it is largely functional in its semantics. Your definition of FP seems ahistorically narrow...
8/18
2016
@kevinchen but we've only exercised it a little. And good-in-theory support is not the same as a good library of easing animations. @kevinchen yes, in principle, by using time as another input. Incremental has built in support for efficient handling of time as an input. @samth sure, but it's not unavoidable. There are tradeoffs organizations can make around how closely they coordinate. It needn't degenerate. @alexott_en Sometimes, that's technical debt in disguise. Working around poorly documented and designed systems requires more talking. @samth say more! Is your point that coordination costs show you down? I guess that's an organizational structure problem. @dustineichler and sometimes client and server are two parts of the same program, rolled as a unit, but happen to run on different machines. @dustineichler you don't always want to share, but the option is lovely. @dustineichler Say more! Which concerns do you think are not being appropriately separated?If the answer is down, why? Cultural issues? Technical debt? Insufficient tools?A question to think about it you manage a dev team: has the velocity of individual devs gone up or down as your organization has grown?So concepts and code can be shared between client and server, with everything in OCaml.What's appealing is that Incremental acts as a single mechanism that can be used in client and server to present your data efficiently.I'm excited about the improvements coming to incr_dom. It's turning into a really lovely way to write high performance, data intensive apps.
8/17
2016
@PLT_cheater It's still very young...I expect the documentation will improve over time.There's also a portable process management library buried in there, which certainly has lots of uses.We wrote and released it to make it easier to write cross platform build and install processes, but it surely has other uses.This is a fun project: a portable s-expression based shell: https://t.co/NvH52bLv1H
8/15
2016
@jakedonham Yeah, that was my real question. If is just an example, but how do you make that choice in other cases where it matters?
8/14
2016
This kind of thing is important for controlling performance, and I wonder how "transparently reactive" languages deal with this..@ShriramKMurthi In particular, keep both "then" and "else" branches alive, vs, keep just the one currently selected by the condition..@ShriramKMurthi Looking at some old FrTime papers, I wonder: how does FrTime choose between the possible implementations of "if"?
8/13
2016
Why "too much math is driving down diversity" is a flawed argument: https://t.co/RxnLWtDNkU
Retweeted by Yaron MinskyXML is a classic political compromise: it balances the needs of man and machine by being equally unreadable to both.
Retweeted by Yaron Minsky @jhwoodyatt @yminsky I quite like this line, "Asking questions about the earnings on an earnings call is real comedy"
Retweeted by Yaron MinskyHe's both hilarious and deeply well informed. It's really unsettling.The bit about "fancy labor capital" is priceless...I hate @matt_levine because every day he writes more good stuff then I have time to read. Case in point: https://t.co/elVBAgVphx
8/12
2016
@jordwalke Yup. https://t.co/pCmhOmq3JzIn particular, it avoided the need to write out own virtual-dom, as Elm has now done.I think this simplifies the type discipline for end users compared to Elm's approach, and is also easier to implement.Fun with open variants: Experimenting with them as a way to have a functional virtual-dom, without needing a type parameter for the vdom.What some known alternatives to evaluation driven effects? All I know of are monadic effects and algebraic effects.
Retweeted by Yaron Minsky
8/11
2016
Soon, we'll have our open source packages building with Jenga. Important steps towards making Jenga useful outside of Jane Street's walls.We're now starting to cut releases that include our jbuild files, which are used by Jenga to describe the build. https://t.co/RJ0ItseihX
8/7
2016
@DanielleSucher @aphyr @danluu looks like some interesting stuff...Slick elastic collision sim from scratch by a teacher in #bootstrapmodeling. @Bootstrapworld @STEMteachersNYC https://t.co/bEkZu5OBea
Retweeted by Yaron Minsky
8/5
2016
[plant facts!!] bananas are technically berries almonds are seeds avocados are mammals most cucumbers are haunted potatoes aren't even real
Retweeted by Yaron Minsky
8/4
2016
@ballingt @arntzenius either behavior on its own wounds be fine. But the mixing is unnerving. @ballingt @arntzenius the way in which the semantics depends on what's in scope makes the behavior hard to predict. @ballingt @arntzenius the issue is that sometimes the for loop assigns to an existing variable, sometimes it creates a fresh one.
8/1
2016
@mvanier42 I guess, but I still think nearly every dynamic language whose semantics are not essentially those of scheme is mistaken...I showed this to my 14yo son, and he was appropriately appalled. I feel like I've raised him well. https://t.co/PyluZZNaP8
7/30
2016
Spacetime has been merged to OCaml 4.04! More to do, but it's a big step towards an excellent allocation profiler. https://t.co/vuXdX1lmfv
7/29
2016
@czaplic @vdwxv indeed, the whole idea of lists as a privileged data type seems silly in retrospect. @vdwxv better question: why did Haskell move away from the ML syntax? They expected list operations to be more common than types. Misjudged.
Retweeted by Yaron Minsky @ezyang @mvanier42 @kamatsu8 but what should the next Haskell be? @mvanier42 @yminsky @kamatsu8 The next ML should have an effects system.
Retweeted by Yaron Minsky @kamatsu8 @ezyang So, I take it you're not a believer in "the next Haskell should be strict and the next ML should be pure"... @kamatsu8 @ezyang do you think modular implicits are also a mistake? Asking for a friend. https://t.co/SRqZRMVcjq
7/28
2016
@DanielleSucher @mnnakamura And the weak polymorphism issue I think must be there in Scala, since it has uncontrolled effects, like OCaml. @DanielleSucher @mnnakamura I suspect it's the same in both: replacing <expr> with (fun x -> <expr> x). @b0rk @DanielleSucher This covers it from an OCaml perspective: https://t.co/TWark4ZpSu
7/24
2016
Yes! GADTs make things possible and typesafe! Change runtime errors into compile time errors, love GADTs @_chenglou https://t.co/WJUQxpmwIg
Retweeted by Yaron Minsky
7/23
2016
@ShriramKMurthi @PyretLang yeah, we ran into similar problems with Real World OCaml and github.Glad Angstrom's finally released. OCaml finally gets parser combinators that play nice with monadic concurrency libs https://t.co/ngSn45u6qw
Retweeted by Yaron Minsky @dustineichler totally ad hoc, sadly. @boothead Scratch first, when they were 5 or 6. More serious languages at 8 or 9. @yminsky the only common denominator is Scratch!Currently on track to teaching completely disjoint sets of PLs to my 3 kids: Racket, OCaml, openscad, JavaScript, Pyret, Lua, Elm... @PyretLang @github Thanks! I'm really excited to try out Pyret with my 9 year old daughter! @PyretLang I understand the reasons, but there's got to be a better way! I just can't give https://t.co/oQn5TIdTGa full access to my data. @PyretLang looks awesome, but the permissions required for saving to Google Drive are disturbing... https://t.co/lsvCv5XyRF
7/22
2016
If after all we live in a simulation, I bet it's an emacs plugin. https://t.co/XtuUIx9vfE
Retweeted by Yaron MinskyComing soon, courtesy of one of our interns: the ability to write emacs 25 extensions in OCaml. This is going to be so much fun...As promised: What Template Haskell gets wrong and Racket gets right https://t.co/Phak755NWk
Retweeted by Yaron Minsky
7/18
2016
At "Incremental DOM, or, How to Build Reactive Web UIs Without Really Trying" OCaml meetup in SF, pretty excited! https://t.co/zJOAoNF7z7
Retweeted by Yaron MinskySometimes duplicating things, either code or data, can significantly simplifies a system. DRY isn't absolute.
Retweeted by Yaron MinskyReading paxos made moderately complex again. It's really a lovely paper. Clear, no nonsense, and useful for implementers.
7/14
2016
@DanielleSucher bind has the advantage of reminding you of the analogy with let binding. That said, you have a point about flatmap.It finally happened; I had to read https://t.co/ZX6dvGkj2N to remember the syntax for objects cos its been so long since I used them...
Retweeted by Yaron MinskyAfter thirteen years of teaching outstanding math students online, Art of Problem Solving is expanding to... https://t.co/HyzXeFea1f
Retweeted by Yaron Minsky
7/13
2016
@lvicentesanchez modular implicits is not particularly close to being ready. @kragen for sure. But I suspect that, with well chosen abstractions, this is required less often than one might imagine.A key advantage is that you can interest good people in solving high class problems. Less so for low class problems.Low: hand optimizing your UI code. High 1: writing efficient algorithms for building declarative UIs High 2: writing declarative UI logic.
7/12
2016
X

0