Sign in with Twitter

Username:

Occasional OCaml programmer

129 Following   4,480 Followers   3,845 Tweets

Joined Twitter 6/21/09


@zachweed And it's fixed on the blog too.
10/20
2017
@samth @plragde The general purpose mental faculty that I think strong programmers excel in isn't problem solving,… https://t.co/HFlm7wDjEs
10/19
2017
@zachweed The link is here, and I'll see if I can get the page fixed. https://t.co/jSgrQx6Bm5
10/18
2017
@chris__martin @jacobmcarthur Fair, but I do worry that people use specialized terminology more than is helpful. Bu… https://t.co/h0VXWHEB2l @chris__martin @jacobmcarthur And the ability to explain new concepts in plain language is an important and too rare skill. @chris__martin @jacobmcarthur One of the great joys of human language is that this is possible, though typically not with a single word.
10/17
2017
@yminsky this is (in the past few years) Michael Clarkson's class. Probably one of the best (FP) courses in the country right now :-)
Retweeted by Yaron MinskyThis is cool: a list of OCaml projects from Cornell's CS 3110. HW simulator, VCS, extensible text editor... https://t.co/3UpClExIir @leonidasfromxiv Thanks. We'll take a look.
10/16
2017
@gtrakGT We haven't yet published our internal stuff, and a lot of it wouldn't be of general use. I expect we'll re… https://t.co/WdDJyS8cAa @afischer15 Neither: there's an FFI that allows OCaml code to call elisp functions and vice versa. That's why there… https://t.co/Ot5MrYsgeiI also want to go to NOPE now...This is such a great post. I don't know a ton about approximate computing, but I love the style of analysis. https://t.co/QrH5aqJTz2
10/14
2017
@ChiefScientist Ecaml @ChiefScientist And you can read the code on GitHub for more details. @ChiefScientist I don't know the details, but they're linked together. Some tricks we're required to get the GCs to play nicely together.And it's only getting better. Or worse, depending on your perspective.I suppose a little of both is right.Honestly, I'm not sure whether to be proud or embarrassed... @das_kube Sadly, no. To my knowledge, there no good FFI for vim.Did you ever want a nice typeful API showing you how a point works in Emacs? Here it is. https://t.co/nDgFEAUw1VIt's kinda crazy. Look at how detailed the Emacs API is. https://t.co/cX6OpvaFbb @superbobry FFI. It's Emacs 25 and up, only.I don't really know where this ends up, but it's an interesting direction.It feels like we're building an odd, OCaml beast, wearing an Emacs-shaped suit.Things also get faster. Emacs is actually a very fast editor, if you don't load any of that pesky elisp.Testability improves, as does the reliability of the code, because, you know, types.This makes it way easier for our engineers to extend the editor. And the more we move stuff into OCaml, the better it gets.Our editor world has gotten pretty weird. We now have a system for programming Emacs extensions in OCaml, and it's taking over.
10/13
2017
@theindigamer @mdavidallen That's a cool idea.Watch this. Be outraged. Then remember he’s just an official saying it out loud. @ShaunKing https://t.co/BuWKSUWrnt
Retweeted by Yaron Minsky
10/12
2017
@ChiefScientist @JaneStreetNYC That's not our Twitter handle fwiw. @mdavidallen 4 and 7 are doable if you give up on 3... @mdavidallen Or a plot. Any kind of data visualization. @adelbertchang Also the concrete syntax for capturing output of commands and marking off sections of mark-up. So sa… https://t.co/LKQ0OY5xmq @adelbertchang Yeah, we expect to use pandoc as part of our solution as well. @_wilfredh And I do hope we can open source the result. @_wilfredh Org-mode/org-babel also scores pretty well. It gets 7 and 4 but misses 3. @_wilfredh Scribble isn't cross language, sadly. Good code integration with good cross language support is hard.Jeez, threading properly in Twitter is unnecessarily hard.And your web backend can convert that into a pretty picture.So, just dump a textual representation of the data to be graphed into the file. Then the data can be diffed, and reviewed.Here's a nice small idea that came up: how do you make graphical output work. You want it to be diffable and reviewable, after all.And each document is like a single toplevel session. But other than that, it should do the job.It's OCaml specific, using extension points for embedding markup, and expect tests for capturing output.I think we'll soon have a system that does everything except 4 and 7.7. Support for threading together multiple sessions, possibly in different languages, in the same document.6. Good syntax highlighting and navigation for the prose, and a pleasant syntax to go with it.5. Ability to emit to multiple backends: jupyter notebooks, web pages, late, etc4. Cross language support for all of the above7. Support for threading together multiple sessions, possibly in different languages, in the same document.6. Good syntax highlighting and navigation for the prose, and a pleasant syntax to go with it.3. Good integration with code related tools: build systems, indenters, auto complete, type throwback, syntax highlighting2. Diffability, so the end result works well in code review workflows1. Linear readability, so prose, code, and output is unified in one fileThings I want from a code oriented document authoring system:Join us on October 24th at Jane Street for @yminsky's talk on type systems! https://t.co/O9WTwLLFOy
Retweeted by Yaron Minsky
10/11
2017
@R_Thaler Thaler “I do think the film career probably was important to the awarding of this prize. I was disappoint… https://t.co/GODDfHSw95
Retweeted by Yaron Minsky
10/9
2017
@howardabrams @channingwalton Hmm. Looks defunct, actually. https://t.co/iWDk6BP52E @howardabrams @channingwalton Looks like there's something here... https://t.co/ERQOwy9QCx @howardabrams @channingwalton Our current org-to-doc workflows go through Pandoc, which runs way faster. I should c… https://t.co/xla84ctpuj @howardabrams @channingwalton I worry a little about the performance of that. Doesn't org itself take seconds to load? @howardabrams @channingwalton Because the in-editor support seems nice, but to make this scale for the use cases I… https://t.co/h2qRnt40zS @howardabrams @channingwalton What kind of tools are there to hook such documents into a build system?
10/8
2017
@philtor Not necessarily, but London is ideal. @krismicinski And does it support ide like features for those languages? It's sadly kind of a high bar. @krismicinski Does it support non-Racket languages? @samderbyshire Anyway, seems like we're not coming to agreement here, so it seems like time to stop. @samderbyshire I've heard this from actual serious working topologists. @samderbyshire This was the standard view of serious mathematicians in the 90s. Calling it nonsense doesn't make you right.Type throwback, build system integration, auto-completion.It does sound like org mode is close to what I want here. But it needs better integration at the code level. @samderbyshire And again: I do like monads and think they're useful. @samderbyshire Which is to say: "it's mathematically deep! It's category theory!" makes no sense to me. There's ton… https://t.co/X6rfdGjPO8 @samderbyshire A little bit like nonstandard analysis. @samderbyshire But most of the time, that structure doesn't have much juice. For decades, category theory existed w… https://t.co/dH2ruPGl6j @samderbyshire I'm mostly agreeing with your point about pointless generality. Category theory identifies real general purpose structure. @samderbyshire I'm not saying that category theory invents time of silly names. Just that it gives you clear names… https://t.co/QQW9bMEOwt @samderbyshire It obviously has some relevance, but it's not obvious that is especially useful, except for in a few corners.
10/6
2017

0