Topic: Fuck dance, let's code

[Sebastian Cobb]

Oh and for added madness, there's also something called Next JS which executes the front-end Javascript on a server to do the rendering then sends the DOM to the browser.

There was talk of using that in our last job for connected televisions with not much ram. The front end lads used to use shitloads of javascript libraries that would exhaust the resources on their devices then beg us to up the image compression on our backend image renderers and to turn off json pretty printing to save literal bytes of white-space.

[evilcommiedictator]

Java was free hence Unis loved it as they didn't have to pay for Visual Studio licenses, twas all the rage until we figured out it wasn't

[imitationleather]

As if Fatima would ever want to hang out with you fucking nerds.

[Sebastian Cobb]

Java was free hence Unis loved it as they didn't have to pay for Visual Studio licenses, twas all the rage until we figured out it wasn't

And now Oracle are going to strangle it with licensing, another part of Sun's legacy in tatters. Just like their database product, you'd have to have your head examined to build new stuff in it.

[evilcommiedictator]

And now Oracle are going to strangle it with licensing. Just like their database product, you'd have to have your head examined to build new stuff in it.

The Uni I work for does all its internal datawarehousing in Oracle. Guess who is upping the price and going to lose a customer to the M$ cloud?

[touchingcloth]

Java was free hence Unis loved it as they didn't have to pay for Visual Studio licenses, twas all the rage until we figured out it wasn't

Yeah, there are other free languages of course but I think the Java attraction was that the University of Kent had developed (or funded) a free IDE in the form of BlueJ, so students could learn how to work with an IDE without, as you say, the Visual Studio license fees.

I imagine there's way more and way better free development environments now, so students must be having a whale of a time.

[Blumf]

I thought the 'assembly language of the net' JS stuff was just supposed to be a tight subset of JS that wouldn't be particularly readable, but the browsers could neatly optimise and run. Then that's all been overtaken by wasm.

None of which stops idiots dragging in massive frameworks just to call a single, simple function that already exists in the language anyway.

[Sebastian Cobb]

The Uni I work for does all its internal datawarehousing in Oracle. Guess who is upping the price and going to lose a customer to the M$ cloud?

That's basically what they rely on now isn't it? Asking for a bit less than the cost of migrating to something cheaper and in most cases much easier to use.

[Sebastian Cobb]

Yeah, there are other free languages of course but I think the Java attraction was that the University of Kent had developed (or funded) a free IDE in the form of BlueJ, so students could learn how to work with an IDE without, as you say, the Visual Studio license fees.

I imagine there's way more and way better free development environments now, so students must be having a whale of a time.

Ahh Blue J. I recently spotted it and the book it came with in a box of books when looking for something. I suppose I've missed my opportunity to recoup  some of the cost by selling it to some first years.

[touchingcloth]

I can't for the life of me remember what we learned C# in. I can't imagine the uni would have paid for VS, but I'm almost certain whatever it was I used had syntax highlighting for the code, and IntelliSense seems to be one of the selling points of paid VS versus free editors. Maybe I had a cracked version, or MS were giving universities free licences to get the undergrads hooked.

[Sebastian Cobb]

Isn't there a basic community edition or something? I didn't do much at all but we had a soft web module from a lecturer who was more about web trends than anything else and we had a couple of modules dicking about in C# by dragging forms about in VS, it wasn't really coding, but it wasn't wsyiwyg either as you had to code the actions. It was very surface level and I didn't learn much.

[Zetetic]

The point of about transpiling is that it's perfectly possible to build languages with far more pleasant syntax and (for many people) more tolerable paradigms (like class-based inheritance) on top of JS, and the performance characteristics of JS in Chrome and Firefox (as the two platforms now are, pretty much, or their underlying JS  engines if you prefer) feel fairly known... so why not do so?

Biggest issue that we've not really seen anything (other than boring bits like type-systems) gain huge amounts of traction.

Coffeescript, a bit, in the Ruby on Rails world for a bit, which made some sense.

[Zetetic]

Interesting to compare with R, where a certain class systems have gained prominence for certain purposes and tidyverse has come to dominate syntax (esp magrittr pipe) and supporting conventions (often somewhat at odds with standard library).

[Paul Calf]

The point of about transpiling is that it's perfectly possible to build languages with far more pleasant syntax and (for many people) more tolerable paradigms (like class-based inheritance) on top of JS, and the performance characteristics of JS in Chrome and Firefox (as the two platforms now are, pretty much, or their underlying JS  engines if you prefer) feel fairly known... so why not do so?

Biggest issue that we've not really seen anything (other than boring bits like type-systems) gain huge amounts of traction.

Coffeescript, a bit, in the Ruby on Rails world for a bit, which made some sense.

Because sooner or later you're going to need to debug the transpiled code, which will require a familiarity with the target language.

[touchingcloth]

Isn't there a basic community edition or something? I didn't do much at all but we had a soft web module from a lecturer who was more about web trends than anything else and we had a couple of modules dicking about in C# by dragging forms about in VS, it wasn't really coding, but it wasn't wsyiwyg either as you had to code the actions. It was very surface level and I didn't learn much.

God knows - it was 2007/8 when I last used it, so whatever environments were around at the time are lost to the annals of ancient software history.

I wrote an Othello playing program in C# for my dissertation. Well, I say I wrote it, but the approach was to demo your work to date at the end of the first trimester of final year with the understanding that it wouldn't be fully functional, and then to hand in a write up and source code at the end of the year. But the requirement for the source code was as a hard copy, so I said fuck it and plagiarised a load of stuff I had found on the internet and faked the screenshots I needed.

You read that right; a hard copy.

[Sebastian Cobb]

The point of about transpiling is that it's perfectly possible to build languages with far more pleasant syntax and (for many people) more tolerable paradigms (like class-based inheritance) on top of JS, and the performance characteristics of JS in Chrome and Firefox (as the two platforms now are, pretty much, or their underlying JS  engines if you prefer) feel fairly known... so why not do so?

There's nothing wrong with that in theory, but it also papers over a lot of cracks in what essentially is a broken language and it's not going to be clear how exactly they are doing that until their method of abstraction fails miserably.

[touchingcloth]

Interesting to compare with R, where a certain class systems have gained prominence for certain purposes and tidyverse has come to dominate syntax (esp magrittr pipe) and supporting conventions (often somewhat at odds with standard library).

Yes, R is an odd one in that I don't think I've ever seen instructions for anything which involve using it in its stock form, and it's always "first make sure you have the dplyr library..."

[Zetetic]

Because sooner or later you're going to need to debug the transpiled code, which will require a familiarity with the target language.

Less than a C or Rust programmer has to deal with assembly, in my experience with Coffeescript.

[Sebastian Cobb]

A researcher I know was using R for their statistical work, when R wasn't up to the job they used C++ which seems brutal compared to things like Python which also has great stat packages (and matlab stuff if you need it).

[Sebastian Cobb]

God knows - it was 2007/8 when I last used it, so whatever environments were around at the time are lost to the annals of ancient software history.

I wrote an Othello playing program in C# for my dissertation. Well, I say I wrote it, but the approach was to demo your work to date at the end of the first trimester of final year with the understanding that it wouldn't be fully functional, and then to hand in a write up and source code at the end of the year. But the requirement for the source code was as a hard copy, so I said fuck it and plagiarised a load of stuff I had found on the internet and faked the screenshots I needed.

You read that right; a hard copy.

I did a little engine that would pull mms photo messages out of an email inbox (using different methods of MIME extraction for different networks) and a complementary image site that let people view it. I wanted to do a facebook addon to get them in there and ran out of time, and as I was frantically writing the dissertation in the last couple of weeks before the deadline Facebook added a native feature to send images via MMS anyway. Great.

[touchingcloth]

I've never used Python properly - definitely not for statistical stuff - but if I ever have a need to do stats in anger I'd definitely learn it, because R is maddening for it in a lot of ways. IIRC it won't handle comparison operators for integer versus decimal types, or slightly different date types, so the start of every script I've ever written does a load of conversions before the interesting stuff starts.

[Zetetic]

No, that's nonsense, cheers.

(Well, not completely. But lubridate makes datetime handling entirely sane and comparing like-with-like should be something you're doing conciously.)

[touchingcloth]

It could well be. I've mainly used it for joining different spreadsheets together, and have bad memories of having to do a lot of googling and try a lot of different ways to do "if date A from sheet 1 < date B from sheet 2, put "true" in new column C". So not stats at all, just R as a means to a particular end, and not a means I'd use again for the same purpose.

The comparison with int/dec is something I've misremembered but which happens in SSRS. I know you should be conscious of your types, but for fuck's sake don't give me an error if I try and do 2 * 2.5 rather than 2.0 * 2.5.

[Zetetic]

My issue with Python is that it's a language dominated by its constant failure to get anywhere near its own philosophy.

And I think Guido knows this, which is some consolation, but I really hope it troubles him.

[Sebastian Cobb]

My issue with Javascript is that it's 'too broken to fix' now; people have coded round its shortcomings for so long that if it started behaving like it's supposed to everything else would stop working.

[touchingcloth]

Someone should design a language which has no shortcomings, and then at least we could tell Fatima which one to learn. Don’t bother with C, Python, Pascal, they’re all flawed, just learn C^YBer.

[Zetetic]

The comparison with int/dec is something I've misremembered but which happens in SSRS. I know you should be conscious of your types, but for fuck's sake don't give me an error if I try and do 2 * 2.5 rather than 2.0 * 2.5.

R will complain about integer vs. real stuff in many relevant situations, but you almost always have to go out of your way to get an integer.

An integer literal reads like '2L' rather than '2', for example. ("L" because the language is still stuffed with things from decades past, but most of which - in a tidyverse environment - you'll never have to deal with.)

(Given the alternatives: Much of where I work relies on Stata, and that's a fucking nightmare that I've refused to be involved with. Many of us are trying to nudge Epidemiologists and Public Health Consultants towards R, which seems to involve a few of them coming to terms with the fact that they never knew what they were doing and it wasn't their fault.)

[Zetetic]

Someone should design a language which has no shortcomings

It turns out, for humans, it's Ruby, but they throw you in prison if your say that these days.

[Sebastian Cobb]

Gonna invent a language called GaryScript where classes are called 'Gary's' (COS HE'S CLASS) and functions are called Daz's 'cos he's a tool.

[Paul Calf]

It could well be. I've mainly used it for joining different spreadsheets together, and have bad memories of having to do a lot of googling and try a lot of different ways to do "if date A from sheet 1 < date B from sheet 2, put "true" in new column C".

I'm going to try hard to forget I ever read this.

Gonna invent a language called GaryScript where classes are called 'Gary's' (COS HE'S CLASS) and functions are called Daz's 'cos he's a tool.

https://en.wikipedia.org/wiki/INTERCAL