World: r4wp

Having an array type is probably not as useful as you might think. 
The only real advantage I see is the ability to add optional bound-checking, 
which should protect from a whole class of errors.

Array syntax is a little more "average" programmer friendly than 
allocating memory and user pointers though :-)

But only just a little! :D


@Doc I guessed that! It looks other way for outsider(s) (undecided)

Peter, it is possible to add literal array notation support without 
adding an array type.

Also, even if you can statically allocate memory, you still need 
dynamic allocations too.

Peter, what kind of "array syntax" do you have in mind?

For other purposes arrays are not a replacement for pointers. But 
for the majority of small uses of arrays they take away the need 
for the administration overhead. It is handy and easy.

Is there a Red/System page that states its goals? e.g., Is Red/System 
meant to be bare bones, or do you see adding some higher level elements, 
like D. I ask because D has various array types, and didn't know 
if you planned to implment a few core elements at the lowest level, 
so they can be used in Red/System, and then just exposed to Red.

Goals are stated since the beginning here: http://static.red-lang.org/red-system-specs.html#section-1

Red/System is supposed to remain small in order to be usable as a 
Red dialect for low-level programming. I don't have any plans to 
add higher-level features, because they will overlap or collide with 
Red's own features.

So, it's fine to add Red/System features as long as it helps Red 
language construction or extends the low-level accessing abilitites.

Thanks Doc.

I''ll add an idealistic  array wish page to the Github Wiki.

It's here https://github.com/dockimbel/Red/wiki/Ideal-array-capabilities

Comments appreciated.

Re #7, while I understand the reasoning behind the op names, I prefer 
remove/append/take/insert. I'm OK with other terms, though REBOL's 
choices are nice. I used to write OO stuff with ops named based on 
the data structure, e.g. push/pop for stacks, but Bertrand Meyer's 
argument for Linnaean naming conventions swayed me a long time ago.

@arrayproposal: I really like the index to run from 1 to 10 and not 
from 0 to 9. I suspect this is because I happen to be a human-being 
not a computer. Maybe I could live with this different in Red (1 
- 10) and Red/System (0 - 9), because Red is higher level and Red/System 
being low level, but that might be confusing too.

where does I/O appear in the list of priorities and timetable?

+1 Arnold

For Red or Red/System Graham?

I/O for Red/System is likely to be a library rather than in the language 
itself. If I remember correctly Kaj's C lib provides I/O support.

I/O support for Red is very high on the Red todo list  - basic I/O 
support is essential for Nenad for Nenad to write Red/System V2 (which 
itself is essential for "production ready" Red.

I believe that object! datatype and an improved 'LOAD function are 
pre-cursors to I/O in Red and the reason that i/O isn't top of the 
priority list. (Of ocurse, I may be wrong)

Gregg I will note your suggestion in the wiki.

(Though I suspect that my initial wish is very much a "wish to far").

Arnold, MaxV - The array wish includes the ability for the programmer 
to choose between zero-based and one-based indeces as it allows the 
user to specific the array range.

This really is a mega-wish list. It is very possible that adds far 
too much bulk to Red/System to be added in the language.


At least the list could be the basis for the spec for a Red/System 
lib to provide much of the functionality.

@PeterWood - Red thanks

My first impressions on your proposition:


1) arr[i] is a useless syntactic addition as we already have indexed 
accesses: arr/i


2) #7 and #8 are going way too far from the Red/System application 
domain, it's basically  a series abstraction. Red internal API already 
provides series (the internal API is not yet completed nor formalized 
though), so this is both unneeded and overlapping with Red standard 
library.


What you might not realize is that you already have array-like capabilities 
with Red/System pointers (including structs and c-strings). If you 
want automatic memory management in Red/System, you won't have it, 
low-level programming requires a manual management of memory for 
accuracy and avoiding unnecessary burdens.


The only array-like part that Red/System is really missing right 
now is the literal array-like declarations, which can be achieved 
without a new formal array! type. As I said earlier, adding a array! 
type would only add bound-checking abilities (which is a nice feature 
to have) and provide you with #5 as a side-effect (not very useful 
anyway, as array would be fixed-size).

One of Red/System DSL goals is to be as close as possible to the 
hardware, not as far as possible. This implies that no arbitrary 
abstraction layers will be ever added to Red/System that can't be 
more or less directly mapped onto some hardware features. 


Exceptions exist when some targets do not provide a basic feature 
that others are providing. For example, ARMv5 family do not have 
integer division, nor modulo support, so I had to implement it manually 
in the ARM emitter in assembler:

https://github.com/dockimbel/Red/blob/master/red-system/targets/ARM.r#L372


Another one is the runtime exception system recently added (THROW) 
to Red/System, which also required a specific assembler routine:

https://github.com/dockimbel/Red/blob/master/red-system/targets/ARM.r#L2008

Thanks for the clarificaiton, it's helpful. I keep thinking of Red/System 
as a standalone langauge (because I like it so much).

I like it too and I think we can improve it significantly in 2.0.

Being able to do whatever C does without its outdated syntax and 
bloated toolchains is a dream coming true. ;-)

Graham, if I/O is stopping you, you can have it right now in my extended 
Red binaries

With the new shared library code, I have been able to compile a shared 
library on Linux for the first time

I can compile a Red/System extension with the R3 bridge and load 
it into R3. However, only two of the COMMAND examples work:

http://red.esperconsultancy.nl/Red-REBOL-3/dir?ci=tip&name=examples

HELLO and INCREMENT work, the others segfault

Andreas: thanks a lot for this new contribution! Now that shared 
lib generation support is unlocked for Linux, we'll be able to finally 
start working on the Android full support (amongs other new targets).

According to Git it seems that global variables and on-load/on-unload 
are not yet supported. How limiting is that?

Not very, at the moment

The main effect is that Red/System init code is not run when the 
shared lib is loaded, this affects floating point exceptions flags 
and runtime error catching routines. Not a problem for now.

@Peter I took the liberty of commenting the document here https://github.com/dockimbel/Red/wiki/Ideal-array-capabilities
We could add Nenad's comment on it too. 
It is a giant wishlist indeed :)

@Nenad, "Being able to do whatever C does without its outdated syntax 
and bloated toolchains is a dream coming true. ;-)" it is :)

Peter, you should add an "Author:" field in the wiki page header, 
so that it's not being misinterpreted as being a Red official page.

Oh forgot it, I can edit it directly.

DId you by accident wipe the page Nenad?

(No history of the document available, strange)

I changed the title, so the URL changed too, sorry.

https://github.com/dockimbel/Red/wiki/Proposition-for-ideal-array-capabilities

Just click on [Pages] menu to list all pages from the wiki.

I was afraid I caused this.. thx!

@Kaj do you have a url that takes me directly to these files?