Adding staticIota to typetuple.d

[monarchdodra]

First attempt for a static Iota.

http://d.puremagic.com/issues/show_bug.cgi?id=4085

[ghost]

Why are all of these parameters aliases instead of integrals?

[monarchdodra]

To accept floats, or chars. For typed output if you want ubyte/int/ulong.

[denis-sh]

You do no follow name convention. The name must be lowercased. staticIota, iotaTuple etc.
I also suggest to use my code from iotaTuple at least as base point as currently you type inference is incorrect.

[denis-sh]

And yes, thanks.

[monarchdodra]

You do no follow name convention. The name must be lowercased. staticIota, iotaTuple etc.

I'm unsure... Iota (as it has been named), returns a type, so the convention is CamelCase...? In particular, it returns a TypeTyple. I think you said that TypeTuple is wrongly named? But at this point, I think it is better to stick to existing convention.

I also suggest to use my code from iotaTuple at least as base point as currently you type inference is incorrect.

Thanks a lot. Very smart. I shamelessly copied most of it.

BTW, mine handles Iota('a', 'g') ^^.

[denis-sh]

I'm unsure... Iota (as it has been named), returns a type, so the convention is CamelCase...? In particular, it returns a TypeTyple. I think you said that TypeTuple is wrongly named? But at this point, I think it is better to stick to existing convention.

You are confused by TypeTuple name. The template defines generic tuple. Particualry, it can be type tuple or expression tuple (see std.traits.isExpressionTuple and std.traits.isTypeTuple docs). In your case it is expression tuple. So I strongly recommend to at least look at my unstd.generictuple sourses.

BTW, mine handles Iota('a', 'g') ^^.

Never tought about such application. )

---

@andralex, what about finally introduce undocumented GenericTuple and expressionTuple templates just for internal use in Phobos to stop confusing developers?

[monarchdodra]

Allright. staticIota is correct I think. I'll change it. There's also already a private staticIota in typecons I need to get rid of.

[monarchdodra]

Done. It is now called staticIota. It now also checks the types are scalar. I reused the implementation from std.typecons, which has logarithmic depth now, but I fixed some of its bugs ;)

[monarchdodra]

Rebased. Any further input?

[dnadlinger]

Sorry for not seeing this earlier (I'm pretty much out of the loop until end of August), but I'm not sure whether it makes sense to start duplicating all kinds of functionality that already exists as (at least in principle) CTFE-able runtime function‚ just to have it yield a compile-time tuple instead of a range/array.

Wouldn't it make much more sense to offer something like toExpressionTuple!(iota(a, b)) that could be used for other kinds of array/ranges as well?

There might be compiler performance implications to consider, but in general I think keeping things orthogonal is the way to go.

[monarchdodra]

Hum... toExpressionTuple... that's a good idea actually. I'll investigate, but I'm not actually sure where to start :)

[monarchdodra]

I'll investigate, but I'm not actually sure where to start :)

I got it to work. It's actually trivially easy (easier than writing static iota as a matter of fact...) Excellent idea in any case.