2020-05-07 16:44:04 Ivan Reese:

Given that you normalize all syntax (going by the the ** vs ^ example), I'm curious how you might encode something like APL, which has a bunch of syntax that doesn't have an analog in other languages.

...or something like the language Whitespace, which ignores all non-whitespace characters and assigns meaning to spaces, tabs, and linefeeds.

...or perhaps a language where the syntax will have different meaning depending on context. For instance, where a construct like if can be an expression in some contexts and a statement in others.

2020-05-07 18:10:22 George Campbell:

It reminded me of the IR of compilers like LLVM.

2020-05-08 14:46:27 David Piepgrass:

If you're parsing a language that has features that aren't available in any popular languages, feel free to choose your own representation. I just looked up APL and it appears that many of its features are available in more popular languages, with names that are less terse. No one would be able to guess that

(~R∊R∘.×R)/R←1↓ιR Computes which numbers from 2 to R are prime. X[⍋X+.≠' ';] Sorts a word list stored in matrix X according to word length.

So I would recommend renaming APL operators internally using names inspired by Python, Julia and MATLAB, but this is just a suggestion; how you use Loyc trees is up to you.

It is expected that some names will have a different meaning depending on context, e.g. most languages have overloaded functions, but it has become clear to me that processing Loyc trees is easier when their meaning is not ambiguous. For example, I now use a different form for "tuples" and "lists of things that are internal to a construct". A tuple is written (a, b, c) in LES which means `'tuple(a, b, c) while a list internal to a construct is conventionally denoted #(a, b, c), e.g. #fn(T, Square!T, #(#var(T, x)), x*x) marks its argument list with # and represents the function T Square<T>(T x) => x * x`.

2020-05-08 14:55:59 David Piepgrass:

Whereas LLVM is semantically well-defined, Loyc trees represent syntax and so have no semantics, although if you follow conventions, Loyc trees represent "approximate" semantics, e.g. x+2 means "add two to x" which is approximate since it doesn't indicate what x refers to, how many bits there are in 2, or what to do if the addition overflows the size of x. By the way, x is a Symbol object in my implementation and while there is a global symbol called x and '+, you can also define other symbols called x and '+ so that they could refer to a specific variable and a specific version of the operator. Maybe this says more about Symbol than it does about Loyc trees, but I thought I'd mention it because hiding extra information via the symbol object seems useful.

2020-05-07 19:50:10 Ivan Reese:

Like Brian Eno after making the Windows 3.1 startup chime... working in the small scale should give you a new appreciation for what you can do in the large (say, your next conf talk).

For inspiration, cue up twenty four of these: http://5secondfilms.com

2020-05-07 22:09:20 Yair Chuchem:

I follow strictly <=2:00. I think it would be fair if everyone would?

2020-05-08 04:34:50 Ivan Reese:

I don't want to force people to have to edit or do many takes, so if they err on the side of 2:20 I think that's fine. To me, the spirit is more "closest wins" than "price is right".

My video this week was something like 2:40? (We all make mistakes.) One of these times I'll try to do an action-packed 1:20 or so, just for the fun of it.

2020-05-08 12:47:12 Yair Chuchem:

Ivan Reese from my experience of administering a 10k members FB group and getting many complaints over the years, I figured that strictly following well defined rules is the best way to go (people will not have solid grounds to complain on unfair treatment when their posts are deleted if it’s done the same for everyone). Note that another important facet that helps people understand the logic behind the rules is the meta group, which is similar to the meta channel here.

2020-05-08 12:51:37 Mariano Guerra:

I'm usually around 2:20, I could make it 2 if I skipped or simplified the intro/outrou but it will be too punk rock: 1,2,3,4 demo 😛

2020-05-08 12:52:35 Mariano Guerra:

this week it may even be below 2 minutes

2020-05-08 12:53:08 Yair Chuchem:

Mariano Guerra it’s kind of like Twitter or Instagram - the limitations of the format gets something different out of you, and often that thing is good. People who know that they want this kind of content know where to find it.

2020-05-08 15:29:01 David Piepgrass:

Well, Twitter used to be 140 chars and I'm way happier now that it has been doubled... no one would complain if "2 minute week" had a 2:30 limit instead of 2 minutes. And you're right, if I wasn't aiming to put my first video in "2 minute week", it would have been 19 minutes instead of 9½ haha....... somehow this reminds me of KFC's "Toonie Tuesday, a $2 box which they later raised to $2.99 but continued to call it "Toonie Tuesday"

2020-05-08 16:59:36 Duncan Cragg:

2:59.. 2:59... You know you want it

2020-05-08 17:00:07 Duncan Cragg:

I know, let's allow 2:59 but only on Tuesdays!

2020-05-08 17:05:56 Mariano Guerra:

2:59 but you can post once every 1.49 weeks 😉

2020-05-08 17:37:03 Ivan Reese:

Yair Chuchem — The hard limit is <=2:59. If ~Duncan~ someone posts a 2:59 video, I won't ask them to remove or change it. If they post a "much closer to 3 minutes than 2 minutes" video every week, I'll likely kindly ask them to try to get closer to two minutes flat.

After the 280 character limit came into effect on Twitter, the wisdom that I frequently heard, and came to agree with, was: All the best tweets are still under 140.

I think my "$299" remark is probably the best encapsulation of what I think the spirit should be: https://futureofcoding.slack.com/archives/C0120A3L30R/p1587247044045000?thread_ts=1587245666.042200&cid=C0120A3L30R