|
rotor posted:just use numbers higher than 255?? doesnt seem hard imo there's even 26 letters to use
|
#
?
Jan 31, 2024 04:37
|
|
|
|
| # ? May 20, 2024 06:46 |
|
|
Internet Janitor posted:every graphical operating system has apis for opening a window, flipping a texture to that window, and obtaining keyboard and pointer events they have that already, its called html5
|
|
#
?
Jan 31, 2024 04:49
|
|
|
programming language idea: java, but with html5 literals, and more support for functional programming so that someone could write a popular library that halfass emulates haskell
|
|
#
?
Jan 31, 2024 04:50
|
|
|
fart simpson posted:programming language idea: java, but with html5 literals, and more support for functional programming so that someone could write a popular library that halfass emulates haskell just VMs all the way down
|
|
#
?
Jan 31, 2024 04:56
|
|
|
fart simpson posted:programming language idea: java, but with html5 literals, and more support for functional programming so that someone could write a popular library that halfass emulates haskell don't forget to have the self-hosted compiler as dogshit slow as humanly possible
|
|
#
?
Jan 31, 2024 04:56
|
|
|
Dijkstracula posted:don't forget to have the self-hosted compiler as dogshit slow as humanly possible good catch!
|
|
#
?
Jan 31, 2024 05:09
|
|
|
fart simpson posted:they have that already, its called html5 "html (living standard)", excuse you
|
|
#
?
Jan 31, 2024 05:12
|
|
|
Dijkstracula posted:don't forget to have the self-hosted compiler as dogshit slow as humanly possible it's self-hosting, but only because the compiler was mechanically translated from another, faster language
|
|
#
?
Jan 31, 2024 05:14
|
|
|
fart simpson posted:programming language idea: java, but with html5 literals, and more support for functional programming so that someone could write a popular library that halfass emulates haskell Dijkstracula posted:don't forget to have the self-hosted compiler as dogshit slow as humanly possible here you go http://www.impredicative.com/ur/
|
|
#
?
Jan 31, 2024 05:22
|
|
|
he'll yes
|
|
#
?
Jan 31, 2024 05:23
|
|
|
now this page right here tells me everything i need to know. this is a pure programming language http://www.impredicative.com/ur/tutorial/
|
|
#
?
Jan 31, 2024 05:25
|
|
|
row types seem really cool. I’ve never done any ML but it seems super sweet. over in “doesn’t really check types” land, I’ve used MATLAB’s data tables in vaguely row-type like ways ("""polymorphic""" functions for working with general time series assume there’s a column t; working with intervals assumes there’s a t0 and tf, etc etc). I guess you can do the same thing but actually get some type checking over in TypeScript. Idk how easy it is to get the duck-typing + type checking experience in Python. yeh, I know the ML row polymorphism is better
|
|
#
?
Jan 31, 2024 05:58
|
|
|
even if some sort of low-level cross-platform graphics library is a good idea to build into a stdlib, i do not think that c++ is the lang that should pioneer this
|
|
#
?
Jan 31, 2024 07:20
|
|
|
or just use Skia / whatever wraps skia in your language
|
|
#
?
Jan 31, 2024 08:04
|
|
|
redleader posted:even if some sort of
|
|
#
?
Jan 31, 2024 10:22
|
|
|
Sagacity posted:
|
|
#
?
Jan 31, 2024 10:23
|
|
|
Sagacity posted:speaking of which, did they merge that entire proposed 2d rendering library into the standard library yet nope a couple proponents got real pissy when the objections started rolling in too, almost like they were trying to fly under the radar with their pile of bullshit
|
|
#
?
Jan 31, 2024 12:02
|
|
|
leper khan posted:was the problem that the committee was being kind and not telling the person "lol. lmao. no." amd instead giving constructive feedback on a proposal that would never accept? no, the people submitting the proposal are part of the committee and thought well of by other members of the committee, by and large but it turns out that they’re not the only people whose votes need to be considered
|
|
#
?
Jan 31, 2024 12:03
|
|
|
Internet Janitor posted:it would be nice if the creators of more languages and language ecosystems considered being able to do that sort of poo poo out of the box essential, rather than some kind of esoteric distraction from the real language approach of offering stdio and leaving users to write their own bindings to C libraries if they need anything else no it wouldn’t instead maybe consider that different platforms are different for any number of reasons and the idea that a language or standard library can “smooth over” differences is so wrong-headed as to be beneath consideration I learned recently that someone was proposing adding path handling concepts to C++ and (real talk) given the mess that Common Lisp path handling actually is that’s beyond delusional-but-functional and into involuntary-commitment territory IMO
|
|
#
?
Jan 31, 2024 12:08
|
|
|
Internet Janitor posted:I am emphatically not talking about a "UI framework"; i'm talking about a minimum-viable IO layer like SDL, Raylib, and Allegro, which have been a perfectly acceptable "experience" for building hundreds of thousands of games and applications they really don’t though by and large things built with them are either fully immersive games with zero platform UI (which sucks) or absolute garbage
|
|
#
?
Jan 31, 2024 12:09
|
|
|
Internet Janitor posted:awt and swing look dated and non-native, and are kind of a pain in the rear end to use, but they're functional. with a little effort you can get results comparable in many ways to Qt, but unlike Qt you don't have to run all your code through an insane custom preprocessor all of these but AWT are absolute loving garbage and bbcode doesn’t let me emphasize that nearly enough AWT is only mostly loving garbage because at least it explicitly uses native UI wherever possible, the biggest problems are that it’s up to the app(let) developer to ensure the UI layout matches the platform and that it assumes menus are associated with windows rather than applications
|
|
#
?
Jan 31, 2024 12:14
|
|
|
Internet Janitor posted:at no point have i remotely implied i want to be able to make anything look like it's using native widgets with native styling and while this is awesome for Decker, it absolutely loving sucks for any applications anyone needs to use to get work done or do other non-optional tasks throughout their day I honestly, sincerely wish that the iOS App Store had outright rejected applications from day 1 that used non-platform controls where platform controls existed, and that neither web views nor JavaScriptCore were ever a thing the only other thing I wish with such sincerely about modern platforms is that “consumable” in-app purchases were never a thing, honestly, treating application UI as “branding” has been such an absolute disaster for usability that it’s set the industry back by decades
|
|
#
?
Jan 31, 2024 12:18
|
|
|
eschaton posted:
Common Lisp's path handling is comically complicated by today's standards due to the diversity of file systems it had to support. I don't think it would be nearly as convoluted to design a portable path handling library that intends to support only Windows and Unix paths. I don't care at all about how C++ people screw up their language, though. I only care when their weird tendencies affect the rest of the world (such as when they insist C header files must also be valid C++).
|
|
#
?
Jan 31, 2024 12:19
|
|
|
i am quite convinced that the idea of consistency of look and feel being important is nonsense. it was some 80s idea which took a bad guess at what made computers hard to use for people, and it is clinging on entirely contrary to the ample evidence that it does not matter the least bit if everything looks the same, as long as things by and large look like the thing people expect, and more important *work* the way people expect. i.e. no defense of random clickable text, and absolutely not for doing a custom text control which e.g. breaks ime or keyboard shortcuts, but the button having a different color? does not matter. if it made an appreciable difference there would be a large movement to destyle websites.
|
|
#
?
Jan 31, 2024 12:22
|
|
|
Vanadium posted:I see the appeal of just not doing native ui but whenever I have to use some text edit control that implements yet another completely different, wildly insufficient subset of the behaviors that every Windows text box since '95 supported, it does make me feel like a high level widget toolkit ought not only to ship with the programming language but actually be mandatory to use unless you have a waiver from one of those old times HIG experts such a toolkit ships with your platform of choice and the waiver should be to use something other than it every time a web view pretending to be an app overrides the standard macOS text editing commands because the webshits who wrote the megabytes of JavaScript depleting my battery thought they knew better than the actual experts working on TextKit, I want to use my fists and their faces to generate some expressionist art
|
|
#
?
Jan 31, 2024 12:24
|
|
|
Athas posted:Common Lisp's path handling is comically complicated by today's standards due to the diversity of file systems it had to support. I don't think it would be nearly as convoluted to design a portable path handling library that intends to support only Windows and Unix paths. and thus is bullshit because it assumes Windows and UNIX paths are the only style of paths that will be evermore
|
|
#
?
Jan 31, 2024 12:26
|
|
|
so, like, kindly rank these two people in level of insanity: 1. the person who suggests we create some convenience functions to manipulating paths, handling the only two cases in use today 2. the person designing a new system deciding to define what a path is/how it works in an entirely different way like, you figuring that maybe the filesystem tree will become a general (maybe hyper-) graph any moment now?
|
|
#
?
Jan 31, 2024 12:30
|
|
|
Cybernetic Vermin posted:i am quite convinced that the idea of consistency of look and feel being important is nonsense. I wish the forum rules let me express how I really feel about this statement
|
|
#
?
Jan 31, 2024 12:30
|
|
|
Cybernetic Vermin posted:so, like, kindly rank these two people in level of insanity: there’s a world of difference between “hey let’s create some convenience functions for this codebase” and “hey let’s put path handling in the standard library and force it on all implementations of the language” the filesystem may become any number of things, why constrain it to what!s common today? maybe record-oriented I/O will become a thing again. maybe relational tables will be integrated with filesystem B-trees again. who knows? why standardize something that doesn’t need to be?
|
|
#
?
Jan 31, 2024 12:36
|
|
|
eschaton posted:and thus is bullshit because it assumes Windows and UNIX paths are the only style of paths that will be evermore The only thing that is bullshit is having to support lovely Windows paths at all. I hope nobody will be daft enough to make something even more complicated.
|
|
#
?
Jan 31, 2024 12:39
|
|
|
also quantum computing might happen so we should really leave it open to the implementation whether if statements have deterministic outcomes or not
|
|
#
?
Jan 31, 2024 12:39
|
|
|
in other words, I absolute think 1 is worse than 2 I’ve actually used enough systems that have very different but self-consistent ways to organize storage that I think claiming 2 is somehow crazy—or that UNIX and Windows (which is basically UNIX with marginal differences) have “won” and nothing will ever change, which is isomorphic—is itself the crazy thing content-addressable stores are a thing, for example relational and document databases are pervasive, why shouldn’t they be treated as part of the filesystem on a platform? etc.
|
|
#
?
Jan 31, 2024 12:43
|
|
|
i too have used lots of systems where there's a plethora of hierarchies of storage in use, as that's all systems really. in every single one in the last 10 years though these have been additional to, and basically refinements within, a perfectly normal hierarchical file system though. i will give my own expanded answer to the question of sanity here: 1. perfectly sane, provide some ubiquitous baseline functionality for real users 2. insane, it is fine to build a system that does not have a hierarchical file system if you please, every operation then a file not found, which makes sense 3. the eschaton take of "actually provided io apis should be so abstract and general that any imaginable future means of storage fits within it" is absolutely raving lunacy
|
|
#
?
Jan 31, 2024 12:47
|
|
|
eschaton posted:such a toolkit ships with your platform of choice and the waiver should be to use something other than it are you ok don’t get me wrong I love the passion and the posts but you’re giving that fart in class poppin veins guy vibes. which just wanna be clear I love. read every word. never stop
|
|
#
?
Jan 31, 2024 12:47
|
|
|
"your c++ implementation needs to support this filesystem api" and "your c++ implementation needs to do something meaningful with this api even though it's on a platform that doesn't have anything that even remotely looks like a traditional filesystem" are two different statements, and the first does not actually imply the second even if you imagine that somebody creates a c++ implementation for a platform with a stupid filesystem alternative
|
|
#
?
Jan 31, 2024 12:59
|
|
|
eschaton posted:such a toolkit ships with your platform of choice and the waiver should be to use something other than it Absolutely nobody wants to write and maintain a different implementation of their ui code for each target platform.
|
|
#
?
Jan 31, 2024 13:11
|
|
|
arguably perfect consistency is rapidly arriving as everyone targets and uses the web for everything, just got to ban the legacy alternatives from app stores after that
|
|
#
?
Jan 31, 2024 13:13
|
|
|
Cybernetic Vermin posted:
Firefox's reader view is basically this. Also, having consistent look and feel is very important for accessibility - the wall of custom UI inhibits screenreaders, for instance
|
|
#
?
Jan 31, 2024 13:29
|
|
|
Visions of Valerie posted:Firefox's reader view is basically this. Also, having consistent look and feel is very important for accessibility - the wall of custom UI inhibits screenreaders, for instance right, but not specifically in how things are broadly styled is the thing. i.e. complaining about web apps specifically is dubious, as web stuff properly done (and a lot of it is) often works *better* than "native" applications, in that an awful lot of it follows a really friendly scrollable-text-document metaphor, and a few rather simplistic modes of interaction. and then the *styling*, the precise way they choose to have a button look, does not matter that much. by comparison native applications sure love a swipe or drag which are really troublesome to make accessible in general. e: but to be clear this is a bit devils advocate, just happens to be relatively easy since it is very much the winning side regardless of merits.
|
|
#
?
Jan 31, 2024 13:35
|
|
|
|
| # ? May 20, 2024 06:46 |
|
|
what do you guys think is going to happen to the bespoke platform, which doesn't have a "traditional filesystem", when they skip supporting the c++ filesystem standard? the standard assassins will show up and murder them?
|
|
#
?
Jan 31, 2024 13:48
|
|
