LWN: Comments on "The Perl 6 release"

The Perl 6 release

mathstuf — Thu, 07 Jan 2016 17:03:24 +0000

In my terminal (using Terminus), the difference is 1 pixel: the letter x has a 2-pixel center stem which puts it on the baseline while the other "floats" a single pixel above the baseline due to its shorter stem.

The Perl 6 release

jwarnica — Tue, 05 Jan 2016 06:16:50 +0000

You think the adoption of TeX will *improve*? What over the past 38 years has indicated that TeX will ever be used for anything but obscure technical fields? TeX - even LaTeX - requires that someone be both an author of a work, be capable of understanding typographic concepts, and *care* enough to invest the time, rather than just using, well, anything else. That is an exceedingly small group of super nerds.

The Perl 6 release

nye — Mon, 04 Jan 2016 15:18:08 +0000

>Why not replace the $ with the ₤

Probably because the people designing these keyboards felt that people were more likely to want to write about dollars than about lira.

The Perl 6 release

paulj — Sat, 02 Jan 2016 10:53:38 +0000

Also, note that Scots and Irish Gaelic are very similar (Scots Gaelic coming from Irish Gaelic with Irish missionaries) and are readily mutually comprehensible. Scots spelling is a little different though, but you can usually figure it out.

AFAICT, most Scots outside of the Western highlands and islands (i.e. the vast majority) have no education or understanding of Gaelic at all, note.

The Perl 6 release

paulj — Sat, 02 Jan 2016 10:38:30 +0000

Obviously, this is simplifying things to a staggering extent, but...

A (significant) minority of people in Ireland are "Scots" (i.e. identify with the community/culture 'descended' from the Scots planted over by Elizabeth I - displacing existing people off their land). Now, those 'Scots' came primarily from the low-lands, and were themselves descended from/identfying with the community/culture that came from/aligned with the English culture - displacing the prior Gaelic culture that existed there. Which itself was an import from *Ireland* in the middle-ages, displacing the pre-Christian cultures (e.g. Pictish) that existed north of Hadrian's wall prior - spear-headed by Irish missionaries. That early Irish Christian culture, by repute, originates with the missionary work in Ireland of St Patrick - a Welshman!

(We will gloss over the North-Eastern Islands of Scotland, which descend/identify more from/with Scandinavian Vikings).

The remainder of Ireland do not speak Irish Gaelic in normal course - other than a trivially tiny number. Although it is a compulsory language to learn in school, and some parents even choose to have their children educated in primarily Gaelic-language schools. Rather, they speak English - with a modicum of local slang. And often speak it better than the English. ;) (E.g., the Irish accent at least continues to pronounce "r" - unlike the mush that the southern English accent has turned into ;) ).

In terms of genetics, the earliest Irish (stone age) came from Spain/France, and further waves of immigration of people descending apparently from the Eastern steppes of Europe. Add Scandinavian vikings, French Normas, Spanish armada sailors who were ship-wrecked, French and Spanish soldiers fighting in Ireland as part of wars of English royal succession (which were split on religious lines) and (of course) lots of english, over the ages. A melting pot. :)

The Perl 6 release

epa — Sat, 02 Jan 2016 09:40:42 +0000

As I understand it, Cornish and Breton are Celtic languages (of the same group as Welsh) but they are not 'Gaelic', which is the other main group including Scottish Gaelic and Irish Gaelic.

The Perl 6 release

dskoll — Fri, 01 Jan 2016 01:17:38 +0000

I think your definition of "English" (the language) is not the definition most people use. After I lived in South Africa where my mother tongue was supposedly English, we moved to Newfoundland where they also supposedly speak English. I can guarantee you that the difference between South African English and Newfoundland English is orders of magnitude larger than that between American and UK English.

Just chill out and accept that Americans also speak English, and have a Happy New Year. :)

The Perl 6 release

nix — Fri, 01 Jan 2016 00:53:23 +0000

Linguists say "British English" because it is a distinct set of dialects, useful for disambiguation when comparing to other broad dialectical groups of English-speakers.

btw, outside of England where they speak English it is almost universally *not* migration that was responsible, any more than migration was responsible for the Anglo-Norman injection into Anglo-Saxon. Even within England the old Saxon kingdom boundaries are discernible in sufficiently carefully-done genetic data. The English have not swamped the Scots, Irish or Welsh (though the Picts are more or less interbred into indiscernibility by this point).

(Also: Gallic != Gaelic. Gaelic is a language *family* anyway, encompassing languages as distinct as Irish, Cornish and Breton. The languages are not the same and are not particularly mutually intelligible, particularly not Irish which has been especially churned by history.)

The Perl 6 release

Wol — Thu, 31 Dec 2015 19:22:59 +0000

> Blimey, mate! :) American and English are much closer than Perl 5 and Perl 6, so I think you can forgive those who conflate them.

:-)

So you're Canadian. I dunno whether you're one of them, but a lot of your compatriots get upset when "the rest of the world" assume that "North America" == "USA". It's the same thing here - I'm English, and my language is English. So when people say "English" and mean "American", I get upset.

Oh, and don't say "British English" because outside of England they don't speak English (well, many do, but that's because English migration has swamped the locals ...). In North Wales they speak Welsh, and in South Wales they're mostly Cornish not Welsh. In Scotland they speak Scots or Gaelic. And in Ireland, again they are mostly either Scots or they speak Gallic.

Cheers,
Wol

Word as a Perl 6 IDE

raiph — Thu, 31 Dec 2015 16:35:39 +0000

I thought the General Category metadata for the characters (Initial Punctuation for « and Final Punctuation for ») was determining.

Anyhow, the full rule set is:

* Bracketing characters are defined as any Unicode characters with either bidirectional mirrorings or Ps/Pe/Pi/Pf properties.

* Characters with no corresponding closing character do not qualify as opening brackets.

* If a character is already used in Ps/Pe/Pi/Pf mappings, then any entry in BidiMirroring is ignored.

* For any given Ps character, the next Pe codepoint (in numerical order) is assumed to be its matching character even if that is not what you might guess using left-right symmetry.

Word as a Perl 6 IDE

songmaster — Thu, 31 Dec 2015 15:51:50 +0000

The Perl 6 language is defined by its test suite, although there are additional discussion documents which are not definitive and point to the associated tests. The Unicode tests can be found at https://github.com/perl6/roast/blob/master/S02-lexical-co... and the discussion at http://design.perl6.org/S02.html#Lexical_Conventions . The section on Bracketing Characters provides some discussion that might be helpful here.

Word as a Perl 6 IDE

alonz — Thu, 31 Dec 2015 13:36:17 +0000

Unfortunately, Unicode has declined to differentiate between opening and closing quotes – instead it has characters like "U+00AB left-pointing double angle quotation mark" and "U+00BB right-pointing double angle quotation mark". So just saying “Unicode” here is tantamount to saying “magic”. Or to saying “we're using Unicode the way it should be done (i.e., the American way)”.

The Perl 6 release

mchapman — Thu, 31 Dec 2015 11:10:09 +0000

> Actually, I will mention one more thing: Perl 6 uses ~ for string concatenation? For the love of $DEITY, why?? I don't know any other language that uses ~ in that way.

In TeX, a tilde produces a non-breaking space. That's almost like string concatenation, I guess.

The Perl 6 release

renox — Thu, 31 Dec 2015 10:36:44 +0000

> Actually, I will mention one more thing: Perl 6 uses ~ for string concatenation? For the love of $DEITY, why?? I don't know any other language that uses ~ in that way.

Unless you're trolling, I'd say you haven't thought enough about the subject:
- first the D language also use ~ for string concatenation.

- concatenation is very different from addition, normally you have a + b == b + a but if you use + for string addition this isn't true.

- having different operators for addition and concatenation is very nice for arrays: if a and b are arrays of int, what means a[] + b[]?
That's not obvious if + can mean addition or concatenation, if you have two different operators that's obvious: a[] + b[] is element-wise addition a[] ~ b[] is concatenation of the two arrays.

I haven't looked much about Perl6 but ignoring *broken* long-established informal programming habits is progress..

The Perl 6 release

JanC_ — Thu, 31 Dec 2015 01:59:49 +0000

> (…) I was going crazy trying to find a syntax error.
> Then I realized the keyword was COLOR, not COLOUR.

At least the folks at Sinclair were sensible enough to use INK and PAPER. :)

The Perl 6 release

JanC_ — Thu, 31 Dec 2015 01:47:40 +0000

"~" is <AltGr>+"=" (which gives you a tilde dead key) followed by a <space> on Belgian keyboards. I think "contortion" is not a bad way to describe that.

The Perl 6 release

juliank — Thu, 31 Dec 2015 00:59:09 +0000

Oh, I forgot about that. This and Cartesian products are the only use of × here. For scalar we use •

The Perl 6 release

juliank — Thu, 31 Dec 2015 00:55:05 +0000

I only know the vertically centered dot for scalar multiplication, does Perl 6 define that as well, or does it only define the * and the × that is only used for Cartesian products over here?

The Perl 6 release

juliank — Thu, 31 Dec 2015 00:49:03 +0000

Python 3 is a beautiful language, Perl 6 feels like the creation of a bunch of (compulsive) hoarders.

The Perl 6 release

juliank — Thu, 31 Dec 2015 00:46:07 +0000

At least here I'm Germany, the × is used for Cartesian product, not for multiplication of numbers. In school and university, that is.

The Perl 6 release

Cyberax — Wed, 30 Dec 2015 23:35:19 +0000

Yes, and Wol's keyboard layout is also different. But you can still enter all special ASCII characters without going through contortions with complicated input methods.

The Perl 6 release

raiph — Wed, 30 Dec 2015 22:18:40 +0000

> It looks like they're trying to disentangle numeric operators from string operators.

Hmm. Let me try to disentangle the issues. :)

Perls have always kept numeric and string operators disentangled as a general principle. Perl 6 retains this aspect of earlier Perls.

> (In earlier versions of Perl at least, operators would force-convert their operands to whatever type they'd expect, i.e., "1" + "2" would give you 3 as a result even though the operands are strings.)

To understand the Perl perspective, drop the concept "expect" and instead think "want and need". And temporarily put aside the sometimes inadequate (from Perl's perspective) notion of "string" and non union (this or that) data types in general.

In Perls, if you write the + operator you are explicitly saying that you *want and need* numeric operands for it. (Likewise with other built in operators that look like math operators. If an op looks like a math op then it wants and needs numeric operands.)

Consider the following program foo.pl:

sub MAIN (\number, \string) {
say number + number;
say string ~ string;
}

(I've used a MAIN sub to help illustrate data that is more than one type at the same time. I've used backslashes to tell Perl 6 sigils aren't wanted or needed in the hope this keeps things familiar for those who don't regularly use a Perl.)

Now execute:

perl6 foo.pl 25 25

Are those 25s strings or numbers? One way of looking at things is that they're obviously strings. Another is that they're obviously numbers.

Perl 5 says it doesn't matter -- it'll do its best. It'll try to coerce to get you what you said you want and if it can't it'll fail at run-time. In this case it can get you what you want so the result is two lines, 50 and then 2525.

Perl 6 retains this approach but provides gradual typing that lets a coder tighten things up as little or as much as they want or need:

sub MAIN (Int \number, \string) {
say number + number;
say string ~ string;
}

Now execute:

perl6 foo.pl 25 25

This still works, for the same reasons as for Perl 5: both arguments are strings AND numbers and Perl takes care of any necessary conversion for you to get you what you said you wanted and needed.

But if you call the tightened up MAIN sub like so:

MAIN('foo', 25)

you'll get a compile time error.

And if you do:

perl6 foo.pl foo 25

you'll get the error message:

Usage:
foo.pl <number> <string>

(Well, you're supposed to, but in my pre 6.c Perl 6 it gets the error message wrong.)

The Perl 6 release

raiph — Wed, 30 Dec 2015 21:08:02 +0000

I just briefly dug back in to the study paper (linked in another comment I posted above). Here's the example they included:

public static void main(String[] args) {
String word = "Hello";
String result = new String();

for (int j = word.length() - 1; j >= 0; j--)
result = result + word.charAt(j);

System.out.println(result);
}

Study about brain activity while coding

raiph — Wed, 30 Dec 2015 20:52:08 +0000

A bit of googling gave http://www.huffingtonpost.com/chris-parnin/scientists-beg... which links to the study and says:

> Interestingly, even though there was code that involve mathematical operations, conditionals, and loop iteration, for these particular tasks, programming had less in common with mathematics and more in common with language. Mathematical calculations typically take place in the intraparietal sulcus, mathematical reasoning in the right frontal pole, and logical reasoning in the left frontal pole. These areas were not strongly activated in comprehending source code.

I haven't reread the study recently but here's some notes from my recollection of it:

They used Java code. It's entirely possible that things would be entirely different for, say, Haskell, though I suspect it would not be for most humans despite the elegance of Hindly-Milner typing and equational logic.

It's about small snippets of code, less than a screenful iirc. It's entirely possible that comprehending larger chunks involves different reasoning modalities.

It's about comprehending code, an extension of reading, not writing code. So it's theoretically possible (but would surprise me) that the brain activity profile when writing code similar to the snippets being comprehended is fundamentally different and focuses on math and linear logic capacities instead.

The study was small. Iirc 15 participants or so. Iirc the results were about universal -- all or nearly all 15 showed about the same activations.

The Perl 6 release

raiph — Wed, 30 Dec 2015 20:28:58 +0000

> Perl 6 uses ~ for string concatenation? For the love of $DEITY, why??

One reason, as I said, is the clear visual pun with a piece of string. There are several others but the near universal "oh, that's really nice and instantly memorable" kind of reaction from folk learning Perl 6 is probably the main one.

I've only seen those who already don't like Perl 6 react negatively to use of ~ as a string operator.

> Why not at least pick something some other language uses already, such as + ?

As I explained in my comment, Perl 6 uses math symbols like `+` only for operating on numbers. Perl 6 takes the view that overloading operators to have unrelated semantics based on the types of operands seriously negatively impacts code readability and correctness.

Word as a Perl 6 IDE

raiph — Wed, 30 Dec 2015 19:55:52 +0000

> Quotes could look «like this» (French style) or »like this« (Danish style), or even »like this» (Finnish). In fact there are about dozen of different conventions. How the hell Perl6 distinguishes them?

Unicode.[1]

Ultimately this is about what devs want. Anecdotally, on the broader theme of Unicode code support, I've read comments by several English speaking coders saying code should stick to ASCII and comments by several coders with a native language not well represented by ASCII saying otherwise.

----

Perl 6 incorporates many innovations such as this and only time will tell which innovations work in practice and which don't. Over time it may become more conservative or less in relation to any given thing it's trying.[2]

----

[1] Unicode provides a unified approach to all aspects of digital representation of written human language, including quoting and bracketing. Perl 6 has adopted Unicode for code and hence for determining quote and delimiter characters.

Most programming languages assume a very limited character set for code. Thus pasting quoted text such as your examples in to a program as if they were string literals will yield a compile time (or interpreter) error except perhaps in bizarre contrived circumstances.

Aiui Larry Wall thinks Perl 6 will yield a correct interpretation a significant amount of the time and a compile time error (even if it's a false negative) otherwise except perhaps in bizarre contrived circumstances. I'm not sure if that's true or how locale variation will be handled but I trust they'll work to prove, improve, or remove this feature over the long term.[2]

[2] Perl 6 has a very strong identity system underlying use of the main language, built in and user-defined sub-languages, test suites, compilers, modules, and user code. Part of the point of this (and some other language characteristics and features) is to facilitate relatively frictionless language evolution without breaking backwards compatibility.

The Perl 6 release

bronson — Wed, 30 Dec 2015 19:33:42 +0000

dskoll is obviously complaining about the cognitive load, not the finger load.

The Perl 6 release

rgmoore — Wed, 30 Dec 2015 18:27:41 +0000

Rather than have an operator + which changes behaviour based on the types of its operands, you have to explicitly say whether you want addition or concatenation.

This makes sense because Perl 5 doesn't make a distinction between numbers and strings internally. Everything is just a scalar, so there's no way for the interpreter to know if you want to treat the scalars as numbers or strings unless you make the distinction with different operators.

The Perl 6 release

flussence — Wed, 30 Dec 2015 18:13:33 +0000

If someone's life becomes difficult because the *bit flipping* operator no longer has a keyboard key reserved for it, may I suggest they take a long *break* from using the keyboard?

Some subset of people are, of course, going to complain that the ternary operator is spelled ??/!! instead of ?/:; the same group are busy complaining that × and x look too similar...

The Perl 6 release

magnuson — Wed, 30 Dec 2015 17:06:32 +0000

This makes sense to me for any language (Java in particular) which is so relentlessly object oriented. With enough libraries there is nary a mathematical operator or C like control structure to be found. It's all object1 (subject) does something (function call/verb) to object2 (er.. object ;)

The Perl 6 release

Wol — Wed, 30 Dec 2015 16:02:57 +0000

Keep death off the roads, drive on the pavement ...

Will be understood completely differently depending on whether you are English or American :-)

Cheers,
Wol

The Perl 6 release

dskoll — Wed, 30 Dec 2015 13:01:31 +0000

That's all well and good, but ~ in Perl 5 is the unary bitwise-complement operator. I see that in Perl 6 it's been changed to +^. Yet more baggage and gratuitous differences.

And yes I know, I know, Perl 5 and Perl 6 are different languages. But I'd hazard that over 80% of Perl 6 adopters will be coming with a Perl 5 background, so why make their lives more difficult than necessary?

The Perl 6 release

micka — Wed, 30 Dec 2015 13:01:27 +0000

My language (french) uses the latin alphabet. The keyboards here are not QWERTY but AZERTY. There are maybe 5 letters in different places but nearly any non-letter is at a different place.

The Perl 6 release

niner — Wed, 30 Dec 2015 12:43:39 +0000

I get it, you're a Python fan. There's nothing wrong with that.
What I don't understand is what you're actually doing in a discussion about Perl 6? Why do you feel the need to denigrate a language, that you have never used and have never intended to use? Fells like your attacks tell more about yourself than about your target.

The Perl 6 release

Cyberax — Wed, 30 Dec 2015 12:06:07 +0000

Chinese numerals are still widely used in financial documents (although with a separate "tamper proof" set of symbols). So if an accountant is going to write software in Perl6, then Perl6 must support alternative forms of numbers.

After all, "there's more than one way to screw it up" is Perl motto.

Oh, and we MUST be able to use 下 and 上 instead of < and > for comparison.

The Perl 6 release

rsidd — Wed, 30 Dec 2015 12:04:14 +0000

You're saying Perl 6 uses it for vectors and cartesian products? That would be sensible.

The Perl 6 release

HelloWorld — Wed, 30 Dec 2015 11:47:12 +0000

> × is rarely used in actual mathematics
Yeah, because who needs vector products or cartesian products in “actual mathematics”, right?

The Perl 6 release

HelloWorld — Wed, 30 Dec 2015 11:41:43 +0000

I doubt anybody would write it like that. Arabic digits tend to dominate even in China. The typical way to write, say, today's date is 2015年12月30日 rather than 二〇一五年十二月三十日. And 两 is only used if followed by a counting word, making that even less likely.

The Perl 6 release

Cyberax — Wed, 30 Dec 2015 11:23:05 +0000

I support lots of Python code. It's markedly easier to support than "fun" Ruby or Perl (blergh) projects.

I rarely see "fun" code, but Python code at least usually does not make me want to eat my own eyes and murder the developer who wrote it.

The Perl 6 release

epa — Wed, 30 Dec 2015 11:22:32 +0000

I think this is something which Perl 6 has in common with Perl 5: a preference for monomorphic operators. Rather than have an operator + which changes behaviour based on the types of its operands, you have to explicitly say whether you want addition or concatenation. For the same reason Perl 5 distinguishes between numeric equality == and string equality eq. (That said, I think Perl 6 does introduce more polymorphism in some cases.)