Quantcast
Sign in with Twitter

Username:

Occasional OCaml programmer

88 Following   2,061 Followers   1,121 Tweets

Join Twitter 6/21/09

Reverse Tweets
@vu3rdd Fixed.Some thoughts on how centralizing your version control system can lead to better merges. https://t.co/Gc3uTMTDOA
3/4
2015
@pchiusano I do think adding an infix operator should be though of as expensive. You should weep a little each time you add one.@pchiusano But I do worry about idioms that work well for the initiated, but that one struggles to explain to more casual readers.@pchiusano there's something to this, and yet, by this criterion, I'd prefer Scheme to typed languages, which I don't.
3/3
2015
My first pull request to the OCaml compiler. A daring change indeed. https://t.co/JvQtSFuLVf
3/2
2015
@rnml0 I'm only flipping part 1. Part 2, where core command is, has some more platform specific bits.
3/1
2015
@das_kube The subtyping facilities of polymorphic variants make them complex, but it's the concision that makes them valuable typically.@das_kube The typing discipline around polymorphic variants is error prone, making some typos typecheck that ordinary variants catch.@das_kube Quite the opposite. I would use polymorphic variants even less, now that type directed disambiguation exists.@avsm @natefoster that's probably for the best.
2/28
2015
@hmansell @ndm_haskell You guys rock. Thanks. I'm hoping this can inform some improvements to OCaml.@hmansell @ndm_haskell Makes sense. Is there a summary I can read of how this function affects gc speed?@ndm_haskell do you know the form of the size hint?
2/27
2015
Wrote a little network server in @racketlang today. Very pleasant. May have to port to something else to avoid freaking out coworkers.
Retweeted by Yaron MinskyA key part of any great @icfp_conference submission is a @PhilipWadler-esque title.
Retweeted by Yaron Minsky
2/26
2015
@yminsky I'll take a look@bitbckt link?@bitbckt but the two are connected. Running the major GC is required for collecting the off heap memory.@bitbckt which is essentially to pretend that they have size zero...@bitbckt OCaml has some heuristics for this. Is the normal approach to ignore the size of these off heap objects entirely?@bitbckt the number and size of off-heap objects should influence how fast you push the collector.@bitbckt e.g., if you wrap gmp, then you will have pointers from your heap to the gmp objects in the malloc managed heap.Anyone know a good approach for adjusting the speed of a GC to account for out-of-heap values? What does Java do? Haskell? Dot-net?Reading code is harder than writing code because you have to understand the code if you're reading it.
Retweeted by Yaron Minsky
2/25
2015
@llaisdy Not sure, but I'm hoping to get a first version online in a month or so. There won't be much new content at first though.
2/24
2015
Really nice paper able OCaml's upcoming(?) ad-hoc polymorphism mechanism. http://t.co/fw31aaNeUp http://t.co/MD2VaXR30R
Retweeted by Yaron Minsky
2/23
2015
@marius also, you might want to recharge...@marius right you are. I'm quite surprised...@marius I'm pretty sure that was SPJ. Bob doesn't think Haskell is the world's best anything.Really, you should judge a functional programming language on how much it helps you program with effects.This comes I worry from an aversion to effects. But this is crazy. Effectful programming is of fundamental importance.@yminsky perhaps they mean that it has no aspects that do not function. Or perhaps they are using purely as an emphatic, like 'hella'
Retweeted by Yaron Minsky@igstan I think that's irrelevant, but that's a somewhat contentious point of view, for reasons that escape me.Never know what to say when someone mentions their "purely functional" networking code. Should I explain that sending a packet is an effect?
2/19
2015
If your language makes illegal states unrepresentable, then only criminals will have unrepresentable states.
Retweeted by Yaron Minsky
2/18
2015
@natefoster Possibly, though the timing seems a bit off...
2/16
2015
@yminsky Learning a new language is a great way to pass time during a blizzard.
Retweeted by Yaron MinskyInterestingly, Amazon stats and website hits for Real World OCaml, and # of opam users are up big in the last 6 weeks or so. Not sure why.
2/15
2015
@tianyi linguists are hard core...
2/10
2015
@avsm @kc_srk also take a look at core's Command library. https://t.co/hwJR7Wa67t@lambdahands FWIW, that should be fixed soonish. We're reorganizing Core to not use packed modules, which will resolve the issue.@das_kube Hah! I don't know that those will land in the same release, but OCaml is surely improving fast.@das_kube more improvements to the current GC than a brand new one. But yes, all changes are intended to be upstreamed.Had fun learning from D. Doligez how poor our understanding of the GC was. But both our understanding, and the GC, are improving rapidly.@typeocaml Mostly because I almost never use it...Just started updating text for a refresh of Real World OCaml. Struggling a bit with how to explain type directed constructor disambiguation.Music to my ears:@yminsky on the importance of increasing productivity to solve the problems that matter. https://t.co/397Ejs79lZ
Retweeted by Yaron Minsky
2/9
2015
Rewriting a bad old RPC system from a decade ago, struck by how little I knew about modular design. I know more now, but it's still hard.@yminsky Though we also need to get opam on Windows working smoothly to truly achieve non-embarrassment.Looking forward to a second edition of Real World OCaml, using Core_kernel instead of Core. Non-embarrassing Windows support here we come!
2/6
2015
@ApocalypseHow @CBR Same here. Looking forward to the comic!The patch making Async_kernel depend only on Core_kernel just got through review. Anyone care to volunteer libuv bindings?@krismicinski You might be surprised to learn we're even willing to hire people who have never gone to grad school!A friendly reminder: Jane Street is (still) hiring. Join one of the world's best teams of functional programmers. https://t.co/hkgp7eNk8N@andrejbauer MIT dropped scheme for python years ago. But python, PHP, ruby, and javascript are all eying optional static type systems.@TailCallingDev @mavnn Also, I think it's hard to argue that dependent types cover the same territory as macros.@TailCallingDev @mavnn Macros, like HM types, have been practical since the 70's. Dependent types are promising, but not yet practical.Now that dynamic language folk have been convinced of the necessity of types, it's time to push on better macro systems for typed languages.
2/5
2015
"I don't use GADTs - they're too modern for me." "Weren't they your proposal in the first place?" "Yes." #dsldagstuhl
Retweeted by Yaron MinskySomething there is that doesn't love an abstraction / That wants it leaky
Retweeted by Yaron Minsky
2/3
2015
@hausdorff_space Agreed, though we've found tools to be a worthwhile investment even for a fairly small (400 people, 65 devs) organization.A short post about the importance of tools, and the kinds of dev tools we build at Jane Street https://t.co/T45N3Yh0OF
2/2
2015
@danclien Not really. Maybe I can wrangle a post out of the people teaching it this year.http://t.co/MNhoal7AQH - "Teaching programming is an opportunity to reflect on how unnecessarily hard programming can be"
Retweeted by Yaron Minsky@aphyr we've moved past microservices into nanoservices, where each service gets exactly one URL to serve. We host them in femtocontainers.
Retweeted by Yaron MinskyDon't share code between microservices. Write a new handcrafted HTTP microclient for each microservice. Run them on artisanal microkernels.
Retweeted by Yaron Minsky
1/31
2015
"I saw what he meant", from Michael Jackson. http://t.co/IX1XnjgYNr (section on Brilliance), h/t https://t.co/1vJulaCsUt
Retweeted by Yaron Minsky
1/24
2015
@jakedonham oh, so it's a Haxl clone! Great. Maybe we should make one too....@jakedonham what's with all the food pictures? When i give talks, all I have in my slides is code...@jakedonham Stitch is monadic rather than applicative, yes?@jakedonham putting together EDSLs for assembling simple parsers. Like a generalization of Command: https://t.co/KaQ2JRaAQLAnd you can expect Core_kernel.Applicative to show up on github soon...There's an exciting rush of people starting to use Applicative within Jane Street. I'm surprised by how useful it's turning out.
1/23
2015
The overhead for getting students set up to use OCaml is approaching tolerably low, but we're not quite there yet... http://t.co/rEpZXUJUBS
1/22
2015
@avsm @natefoster Oh, and it looks like one could easily use the Fieldslib macros to help autogenerate lenses from record fields. Nice.@avsm @natefoster awesome. I'll take a look.Is there a good lens library for OCaml? @natefoster
1/21
2015
Kudos to Professor Nate Foster, whose paper was awarded the 2015 Most Influential at Principles of Programming Languages 2015. #popl2015
Retweeted by Yaron MinskyPaper of the Week is back! This week's paper is Xen and the Art of Virtualization https://t.co/IVA7jhuIOP
Retweeted by Yaron Minsky
1/20
2015
it wouldn't shock me to learn we should have all been using OCaml since 1996
Retweeted by Yaron Minsky
1/13
2015
@das_kube indeed. And that's part of the proposal.A proposal to rework (and simplify) OCaml's float-array specialization: http://t.co/QrMjt9TqID
1/10
2015
Got a long list of ex-students… they’ll tell you I’m insane http://t.co/mTOBM7nDrf, thanks @justinesherry
Retweeted by Yaron Minsky
1/7
2015
@jakedonham by Dave Powers, actually. I'm not the only one who contribute to the blog...
1/6
2015
The 13 Virtues of Code Review (with apologies to Ben Franklin) https://t.co/SMCWxWFYLO
1/2
2014
Mirage 2014 review: IPv6, TLS, Irmin, Jitsu and community growth: http://t.co/rjVBZVptO6 Comments: https://t.co/2Sk5kDuIQM
Retweeted by Yaron Minsky@marius My 13 year old still codes in plain text. It's not dead yet...
12/31
2014
@ajaxchai I'm also curious which API changes have bitten you.@ajaxchai Indeed. We're working on an update to the tool chain right now so we can more easily keep it up to date.
12/26
2014
@samth Put another way, do you disagree with my earlier point about pragmatics?@samth That's sensible enough, but do you really think the pedagogy is unchanged by the language?@samth Indeed, imagine teaching htdp in C. Language choice affects the pragmatics of different pedagogical approaches. And pragmatics matter@samth Yes, but language isn't trivial. Different topics (e.g., abstraction and machine model) are easier in some and harder in others.
12/21
2014
@typeocaml Fixed! Brown out in Cambridge.
12/18
2014
@samth Makes sense. If you don't think of hack and flow as ML-like, are they TR-like? And are they pragmatically ML-like?@samth interesting! I would have thought that pragmatically, ML and typed rkt would be closer, with more idioms transportable between them.@samth How would you describe Typed Racket? I think of it as more ML-like than, say, Eiffel-like. But I'm probably thinking about it wrong.Is it accurate to describe hack and flow's type systems as being ML-like?@jeffreysco Given that you've used OCaml on iOS, I'm really curious how you think Swift compares.
12/15
2014
« Prev1234567

0