CMQ: I'm thinking of combining the two "inverse" APLcart pages /task and /what into a single dual-purpose page. What should the URL be? aplcart.info/⍣¯1? /challenge? /inv(erse)? Something else?
@Adám I prefer "challenge",also like things as "train" or "test"
IMO the problem with calling it "inverse" or smth of the sorts is that the URL then doesn't give a clue about what you'll find there, while challenge/task/train/test/... are more clear
@Adám sure, I did realize "train" has that connotation. Maybe "training" or just drop it. Task is shorter than challenge, two nice possibilities... Will sleep over it and let you know in the morning
@Adám - I'm not sure that ⍣¯1 would be a good idea; there are enough old DNSsen out there that will choke on anything but basic Latin-1, and the non-Latin-1 characters will look fugly in URLEncoding.
A related problem is that I'd like a (not too verbose) link from the main page, but how do I phrase it so it doesn't sound like APLcart is asking to be challenged/tested?
If you wanted to allow for "read testing" vs "write testing", use /testme/read (=current /what) and /testme/write (=current /task) - the /read and /write would be EXTRA-PATH-INFO rather than actual paths.
@JeffZeitlin Since there's so little content on the two pages, and they share most code and styling, I thought I'd make them a single two-section page.
@Adám Thanks - I'm just exploring the SE chat capability. (Are you talking about "APL is a programming language developed in the 1960s by Kenneth E. Iverson."?)
room topic changed to The APL Orchard: Learn and teach, questions about both golfing and general coding. See aplwiki.com/wiki/APL_Orchard [apl] [array-manipulation] [j] [k] [tips]
@Marshall I've bookmarked the two APL Seeds lessons. Do you want an APL Wiki article about the series (like APL Cultivation has) so we can create a table of lessons?
@Adám i'd just leave it as "quiz" also, potentially with a hover text. Also food for thought - ⍎ and ⍕ describe what you have to do pretty well
also more headers yay: https://dzaima.github.io/paste#0vZQxa9tAFMd3fYo3tiQhkcAd1KFTEjqHzuJUnZsjks49ydjBGDoLMhQUnKHBQykoe4YucSEd8y30CfoR@t7pbMvuBVpkirG4e/@73/u/dycBgFt/@uFOft1ezwIFrn9UF59pAq/BxDzfXccC5cPe473RHisa3B0Qw5uCe@T0eg50YuLj626wBllYads4yntjYzbClt/AR7VctD3rdZup9uzOO6Qq7Akow8/Z0xdMUD7UV3Os0nPcV3@E9d4tjZr/zLa2ZNt6ok/uxLKzrTgUoZFNfQ6r4y22lQHjJWVsECQDTOp5Rfvr@d1U165bWy4MlCI0avpegW3FdUVTbc7Qqv0lElNhyMOhPtn/kE6ftqU2akIApz7QzSpMy1q51lGanWr@P9azsxRNDZTBa6pg@yFKgWp2M5TD@lth0AzaAlveCHx7Hdf8iYRaryOr@bVGNu6SWXxHrEGucUiC@uoW/EPn7DLN2fhYKYn62yThkWA5BzngiuVSQSpzYHEsRzyCkcjPIRnGuRjEHEIZCZ79dXawV6R9jO…
I'm reading through Hui and Kromberg's recent "APL Since 1978" and in the discussion of ⌺ (stencil) they give the following example:
{⊂⍵}⌺5⊢'abcde'
abc abcd abcde bcde cde
Why is the {⊂⍵} needed over just ⊂? I'm still pretty new to APL but I would naively think that in general {f⍵...
I'm reading through Hui and Kromberg's recent "APL Since 1978" and in the discussion of ⌺ (stencil) they give the following example:
{⊂⍵}⌺5⊢'abcde'
abc abcd abcde bcde cde
Why is the {⊂⍵} needed over just ⊂? I'm still pretty new to APL but I would naively think that in general {f⍵...
@Marshall @all Thinking about a different esolang, I had an idea for a natural extension of ternaries to more complex if statements while maintaining the same syntax and not explicitly using any control structure words. Was hoping someone could first remind me what they look like in BQN, though!
Couldn't find anything in the GitHub by searching ternary/if statement/control structure...
@dzaima Oh? I thought he said he'd implement ternaries during that first 8v8 thing talk. And that would be it for conditionals, because he didn't want explicit control structures to muddy the language...
@AviF.S. The choose operator selects a function from the right operand (which is a list) and applies it to the arguments. It's defined as ◶ ← {𝕨((𝕨𝔽𝕩)⊑𝕘){𝔽}𝕩} but there's no documentation about it yet.
Also, it's bothered me since day 1 :p I've finally got the bravery to ask: Why did you choose doublestruck 'w' & 'x' as opposed to doublestruck 'x' & 'y'?
Is it to ensure that in the monadic case, the element inside is 'x'
@AviF.S. "should be alphabetically ordered" 2 messages down what i linked
@AviF.S. 𝕨/⍵ is definitely annoying (in dzaima/BQN i still refer to 𝕩 as w due to not yet having bothered migrating the names, which is extremely confusing sometimes)
@dzaima I sure try to though! Lately it's been getting quite busy though! I really think we ought to keep a compilation on the wiki called FAQ, where we just copy & paste the name of the questioner + question & answerer + answer every time one happens, so the room is doomed to forever repeat itself on common questions!
Sort of like keeping an organized archive of question/answer from here. Would people be up to helping contribute, if that were started?
@AviF.S. Would probably be better for me to add a BQN FAQ (FABQNQ) to the BQN repository, maybe specifically oriented towards those coming from APL (FABQNQBAPL).
@AviF.S. a lot is discussed here, so either the FAQ is useless due to there rarely being any case of >2 people asking the same thing, or the FAQ would grow gigantic over not that long of a time
@Marshall It's more of a general issue I'd been thinking about in general though of it not being sustainable to answer for only one person at a time. When things are answered on SE, they're easy for all to find...
@dzaima I was thinking it could be established as part of the courtesy/etiquette of the room that when one asks a question, they should add their question & the corresponding answer to the FAQ page
Yeah, you can't just make all knowledge accessible by summarizing it. The nature of a chat is that people will ask questions that have already been answered but aren't so easy to find, and then the people who don't know the answer learn something.
@AviF.S. you wouldn't have a fancy search engine nor multiple equal question overhead to magically match whatever rephrasing you've made easily searchable in the FAQ, so the only alternative is reading all of the FAQ, which wouldn't work when it grows extremely large and would be much more complicated than just asking someone who personally remembers
@dzaima Shoot that's a really good point... I was thinking it'd be split by category and subcategory and we'd try and ensure each subcategory didn't have more than ~10 questions...
But I hadn't really thought about what would happen as it got really big
@Marshall I wouldn't go that far... I've never once even tried to search the chat log for a question. That's without a doubt impossible
This would at least be plausible to find, and it'd be organized with a consistent syntax, so that if it were successful and we accumulated a large database: It could very easily be parsed and incorporated into a more user friendly interface eventually
But it seems to be having a database of >1000 questions/answers (which is what you're suggesting would happen if you think it'd be impossible to find anything) would be an absolutely invaluable resource, regardless of how we decide to use it...
@AviF.S. also, is it really worth the effort? How often would you have a time of 2 people asking exactly the same question, and both bothering to check in with the FAQ?
@AviF.S. and that's really the problem. APLCart serves a single, specific purpose around a single, concrete topic. This would need to encompass all discussion
@AviF.S. SO has millions of users. We have maybe 10 active enough ones. And even SE struggles with duplicate questions (or quite the opposite, has many many new ones that aren't duplicates)
@AviF.S. Yes, but I think the Wikipedia or APL Wiki model works better for organizing information. There's no reason to phrase everything in a question-answer format. There's also no need to put a Q/A in a specific place; important information, or a link to it, goes on every page where it is relevant.
@dzaima I'd still beg to differ as it'd only be an issue of a good interface like APLCart's
And I, for one, would find it really helpful even in its current format as a compilation of all the answered questions here, rather than skimming through for the golden nuggets of info
@AviF.S. it'd certainly require a google-like search engine, as linking the correct thing for "why does BQN use w/x instead of x/y" with w/x swapped, 𝕨/𝕩/𝕎/𝕏, many many different possible rephrasings (including no "instead of x/y", making matching pretty much impossible as it'd match everything BQN-related), etc. would be extremely hard
Also, one last thing: It's less about the duplicates, which I acknowledge are rare, and more about the vast amounts of valuable information that are forever lost into the void, which seem to me a shame
That's what bothers me personally. I could care less about answering something twice + it's you guys that do it :p
Correct. But 1. It'd certainly be far easier than searching the transcript. 2. At least we'd have the database so that solving that problem could even be a possibility. 3. Given the rapid improvement of technology these days, it's not implausible that computers would be that much better at NLP given some time.
A good way to do this that wouldn't be too not-worth-the-effort would be a way to mark chat messages as important other than stars, and something like the chat's "highlights" mode to view that. It would still accumulate like crazy though, and unfortunately we can't edit SE chat's code
I do think collecting questions and answers would be probably the least effort-intensive part of the process. It takes maybe 20 seconds to copy and paste the exchange, compared to probably several minutes to categorize and deduplicate it. Merging two similar questions would be much harder, and I expect you'd also have the problem that a lot of answers are pretty jumbled; editing for readability would take a while.
@Adám Since you're involved in TopAnswers/APL, perhaps you can advocate for their chat system to have a few more ways to archive messages? Same for anyone involved in Codidact, if that's still holding promise!
@Marshall But that wouldn't be the job of the copy/paster. The job of those asking questions would just be to add the Q/A in the right category, or make a new one if necessary. Later, on our own time of our own volition, we could voluntarily try to merge and such. But that's not the purpose of first compiling something of any meaning
Also, not sure why you say jumbled/edit for readability. Aren't we all competent enough to use a simple, consistent bullet point-like syntax?
@all Boy! This is a much harder sell than I anticipated! Shall I just give in, or are we making headway/useful discussion?
@AviF.S. if you want a question answerer, you need to have information that people would actually be able to search for, which is hard. If you want a place for "vast amounts of valuable information" you need a way to store conversations, which doesn't care about duplicates nor searching. The two need very different infrastructure
@AviF.S. Then you're asking people to do more work in documenting their question than ask it. They won't do it. Answers are often jumbled because of multiple people answering at once, or asking for clarification about the question.
@AviF.S. No, finding or making the appropriate category, and figuring out whether the question is already there or overlaps with other questions, is difficult.
@dzaima Just felt bad since I answered Marshall's before yours. But his was easier...
@dzaima I think I'm proposing something in between. I'm not proposing adding any functionality right now/yet to make it searchable. Just a collection of Q/A copy/pasted. However, I'm also not proposing keeping whole discussions. Again, just non-searchable, minimal-effort cmd-c-v in a category. Everything else would come later if we accumulate something
@Marshall ??? On the Wiki, the categories are listed at the top as hyperlinks. So no looking through, you just read the 'table of contents'
(my evaluation of costs/gains of a Q/A FAQ: costs - need to search through the FAQ before asking a question, probably >95% chance of not finding anything; the need to add to the list upon getting any question answered; gains - the <5% of the time someone actually bothers to search and finds something, probably still leaving some parts unanswered, leading to a question still being asked; ability to read some tidbits of interesting information if reading all of chat is too much work)
@AviF.S. you definitely need to decide on the precise scope of such a thing before much can be thought about it
Just Q/A in appropriate categories. Not worrying about duplicates or alternate phrasings yet. Just compiling a categorized list first. Raw enough that it can be parsed into a better interface. Simple enough that it doesn't take loads of effort. Readable enough that one can still get value from skimming as is
@dzaima I'm not sure I see how so. The important part is the info which can be parsed. From there the format can be changed at will
(if you can barely search chat with all it's overhead of messages and already knowing what exactly you want to search for, i doubt any search interface would ever work unless you involve much AI-powered language parsing, which..)
I'll reiterate though, more than happy to drop it, just continuing to see if we reach some resolution but totally not necessary
@dzaima Perhaps I don't know enough about this sort of stuff, but I still don't see it. Eg. @Adám took lists compiled of idioms from ages ago, like FinnAPL, and parsed them into something searchable
@dzaima And unclear on the constant emphasis on how difficult it'd be to search through. Sure it wouldn't be easy. But our only alternative is the transcript. Relative to that it's a piece of cake to find something!
And for now it's not intended to be searchable, just to be read so that one can constantly learn different bits, without having to read pages and pages of the Orchard
@AviF.S. it still often takes a couple attempts of searching with different terms in it (it's got a lot of synonyms too), and it at least has some set-in-stone terminology, whereas this would potentially talk about barely created things that definitely can be named in many, many ways.
I'd think that troubleshooting would have far more developed and consistent terminology than something as conceptually complicated as 'get the sum of a cell and its corner neighbors'
@AviF.S. BQN, for example. Many things about it are still unnamed (e.g. my use of "immediate operators" is just something i came up with to have a variable name in Java)
If I knew above 'get the sum of a cell and its orthogonal neighbors' was better stated as 'stencil ...' then I likely wouldn't have the problem in the first place
@AviF.S. because you're suggesting to not worry about synonyms at first (adding many many synonyms later or writing them when adding any question would be extremely annoying)
Well, why not give it a shot and start a draft page. No one has to contribute, but if people ever get inspired, they can copy/paste a few. We/I might, in the very worst case, learn something from the firsthand experience of trying to put together such a monster
No one has to participate though...
@dzaima Correct. Which is what Adám did with FinnAPL and loads of other pre-compiled lists. Didn't say it was easy, just that it shouldn't be our current concern
Because then no one would ever contribute anything, if they were expected to do a bunch of synonym/other work on top of it!
@AviF.S. it already had the synonyms "diamondquad quaddiamond tessellate tile cellularautomaton gameoflife vanneumann moore neighbourhoods neighborhoods", and it's one built-in. Imagine thinking of all the alternatives for questions
I feel like the answer should be obvious (as if I've already learned and forgotten it), but how would one take X elements of Y but in a way that Y repeats?
@dzaima Not sure, it just doesn't load. And when I interrupt it, it gives loads of errors so I don't super feel like learning my way around it right now... Sorry, \me being lazy
@AviF.S. weird, 1) it shouldn't be a problem with APLCart, 2) shouldn't happen with localhost: even if there's no server. Does anything happen differently if you just don't start any server?
@AviF.S. if you don't start the server, does localhost:8080 still take a while loading, or does it instantly fail? (also, is python, when started, reporting any accesses?)
(also, unless it's something blocking local requests, maybe worth trying different ports? i.e. python -m SimpleHTTPServer 5432 and localhost:5432/127.0.0.1:5432/any other ip in ifconfig |grep inet)
I'm reading through Hui and Kromberg's recent "APL Since 1978" and in the discussion of ⌺ (stencil) they give the following example:
{⊂⍵}⌺5⊢'abcde'
abc abcd abcde bcde cde
Why is the {⊂⍵} needed over just ⊂? I'm still pretty new to APL but I would naively think that in general {f⍵...
