On Fri, Jul 10, 2009 at 12:42 AM, wrote:

On Thu, 9 Jul 2009, roconnor@theorem.ca wrote:

...

You can use by lib without worrying about the CIE.  You can use my library without ever importing or using the word CIE.  However, the CIE stuff is there for those who need it.

Perhaps I (maybe with some help) need to make a tutorial on the haskell wiki to try to make it less intimidating.

Okay, I threw together a quick introduction at http://www.haskell.org/haskellwiki/Colour.  Any changes, comments, corrections, and addtions are welcome.  It's a wiki!

The word CIE does occur at all in the document.

I read this and it irks me that opaque is not a monoid homomorphism despite being the natural injection of non-transparent colours into semi-transparent colours with pureColour being the projection back. [Incidentally, you have a typo in pureColour, ac `over` mempty should be ac `over` black or opaque black presumably, or even opaque mempty, which I think was what you were going for, illustrating my point.] It's like defining mappend on Integers as (+) and on Reals as (*); actually, I think this is very close to what is actually happening.

Am Dienstag, 7. Juli 2009 14:42 schrieb Robin Green:

On Fri, 10 Jul 2009 10:44:51 +0200

Wolfgang Jeltsch wrote:

...

PASCAL uses “program”, not “programme”,

The word program (as in computer program) is spelled program in both British and American English.

Probably just because British English took it from American English. It’s similar to the “German” word “Computer”. It’s not native. Best wishes, Wolfgang

Wolfgang Jeltsch writes:

Am Freitag, 10. Juli 2009 05:26 schrieb roconnor@theorem.ca:

...

I find it amazing that you independently chose to spell colour with a `u'. It makes me feel better about my choice.

I have to admit that it makes me unhappy. :-(

Why do we use English for identifiers? Because English is the language of computer science. What English should we use? It’s tempting to say, we should use the original English, which is British English. But we should ask again what is the language of computer science. And the language of computer science is American English.

I don't buy that. And don't forget India.

To my knowledge, most early developments in computer science had their roots in the US.

Really? Manchester Mark I, EDSAC I, EDSAC II? Alan Turing, David Wheeler, Maurice Wilkes? To mention a random selection of early ones (leaving aside Konrad Zuse and colleagues and various Russian pioneers on account of not speaking English).

One consequence of this is that reserved words of programming languages are typically in American English. PASCAL uses “program”,

The use of "program" rather than "programme" in programming was mandated by the IFIP in what I regard as an attempt to act outwith their remit. I've never accepted it.

not “programme”, and BASIC uses “COLOR”, not “COLOUR”.

I'm not sure I would use BASIC as an authority for any aspect of programming language design. Going back to the early developments aspect, a high proportion of early work in functional programming was done in Britain and elsewhere in Europe (at a time when Europeans typically preferred British spellings), so perhaps one should recognise that in choosing identifiers. But anyway, where's the harm in a bit of variety? If someone who prefers British spellings originates a package, why get het up about it if they use them in identifiers? I have to put up with American spellings all over the place, so a few British spellings might even up the balance a bit. -- Jón Fairbairn (British, but with a tendency to identify myself as European)

On Jul 10, 2009, at 8:44 PM, Wolfgang Jeltsch wrote:

Am Freitag, 10. Juli 2009 05:26 schrieb roconnor@theorem.ca:

...

I find it amazing that you independently chose to spell colour with a `u'. It makes me feel better about my choice.

I have to admit that it makes me unhappy. :-(

Why do we use English for identifiers? Because English is the language of computer science. What English should we use? It’s tempting to say, we should use the original English, which is British English. But we should ask again what is the language of computer science. And the language of computer science is American English.

It was possible to adopt such an attitude in the 20th century. But this is the 21st century. We have globalisation, internationalisation, localisation. We have Unicode, so that people are no longer limited to the set of characters that technicians from the USA found tolerable back in 1967. (Note that "Americans" includes French Canadians and lost of people who mainly speak Spanish.) Let's face it, by this argument we'd have to abandon the metric system when writing computer programs. There are good reasons to adopt American speling for a term or English spelling. They include the language that the author is most familiar with, the language the paying customers are most familiar with, the language of the laws that the program has to conform to, any number of things. But there is a way above this issue. Why can't we have BOTH "Colour" AND "Color"? What changes, if any, would it take for Haskell to support bidialectal naming practices (whether it is English -vs- American, insular/Brazilian Portuguese, Norwegian, Arabic dialects, or whatever)? We can create an alias for a type easily enough. We can create an alias for a defined function easily enough. We cannot create an alias for a constructor; there is no way to say data X = A | B Int constructor C = B so that C can be used for pattern matching just like B. We cannot create an alias for a module name; while we can create a module that re-exports everything from some module, there is no way to make them act the same as far as inheritance of dotted names is concerned. SML has similar limitations, but not quite the same. It lets us create alias for types and defined functions. It also lets us create aliases for modules, its structures and functors (more or less). But even SML doesn't allow the definition of aliases for constructors.

To my knowledge, most early developments in computer science had their roots in the US. One consequence of this is that reserved words of programming languages are typically in American English. PASCAL uses “program”, not “programme”, and BASIC uses “COLOR”, not “COLOUR”. So, in my opinion, Haskell color packages should use the identifier Color, not Colour. By the way, I also write my papers and documentation in American English.

Best wishes, Wolfgang (who is neither British nor American) _______________________________________________ Haskell-Cafe mailing list Haskell-Cafe@haskell.org http://www.haskell.org/mailman/listinfo/haskell-cafe

Wolfgang Jeltsch schrieb:

Am Mittwoch, 15. Juli 2009 05:27 schrieben Sie:

...

...

Why do we use English for identifiers? Because English is the language of computer science. What English should we use? It’s tempting to say, we should use the original English, which is British English. But we should ask again what is the language of computer science. And the language of computer science is American English. It was possible to adopt such an attitude in the 20th century. But this is

On Jul 10, 2009, at 8:44 PM, Wolfgang Jeltsch wrote: the 21st century. We have globalisation, internationalisation, localisation. We have Unicode, so that people are no longer limited to the set of characters that technicians from the USA found tolerable back in 1967.

So I should upload a package with German identifiers to Hackage?

I most like identifiers like "getZahl". :-)

Am Samstag, 18. Juli 2009 06:31 schrieben Sie:

On Jul 18, 2009, at 2:35 AM, Wolfgang Jeltsch wrote:

...

So I should upload a package with German identifiers to Hackage?

Sure, why not? The fact that I can't read it is my loss, not your fault, and there will be plenty of other German-reading Haskellers to benefit from it. I've happily worked with programs in French (not large ones (:-)).

I don’t think, it’s a good idea to have German identifiers, since Haskell’s keywords are English. On the other hand, I strongly argue that a library about Bézier curves uses the identifier Bézier, not Bezier. So non-ASCII identifiers are useful even if identifiers are in English. Best wishes, Wolfgang

On Jul 18, 2009, at 9:26 PM, Wolfgang Jeltsch wrote:

I don’t think, it’s a good idea to have German identifiers, since Haskell’s keywords are English.

Put it this way: if Haskell's keywords were in German, do you suppose I would write my Haskell code in anything but English? Does the fact that 'data' is a Latin word mean that some fraction of our Haskell identifiers should be in Latin? Some say ita, some say minime.

On the other hand, I strongly argue that a library about Bézier curves uses the identifier Bézier, not Bezier. So non- ASCII identifiers are useful even if identifiers are in English.

Indeed, you need non-ASCII letters to write English well.