World: r4wp

On Windows or UNIX?

oh I got it. I run the console by double click, not from dos prompt.

:-)

sorry :)

np

/ issue fixed.

Kaj, I won't merge dyn-lib-emitter yet, I will work on it as soon 
as I make the new release to add PIC support and start Mach-O and 
ELF implementations. I want all of them working before merging. I 
would probably also make a new release and bump the revision number 
for the merge.

It's rather frustrating to have that functionality available for 
so long and yet not available

It is available, just from an alternative branch. Nothing stops anyone 
from using it.

It does, you don't get the functionality and fixes in master, and 
it's hard to install

It's fine that #include is parsed as a word! because that way it's 
skipped

You can easily keep your forked dyn-lib-emitter up-to-date with just 
a `git rebase`.

I know there is no file!; I don't need it now, I just need file literals 
to be skipped as well, so parsed as a word!, too, instead of bombing 
out

or rather `git merge origin:master`

I can't use Git

I've fixed the runtime lexer to parse #include as issue!, shouldn't 
change anything for you.

OK, that's even better

I've almost implemented file! completly this morning, but am now 
trying to get the multi-line strings literals recognized by both 
compiler and runtime lexers. A bit harder that it seems at first 
look, but should be ready for tonight anyway.

Great

That's great news Doc! File! is a huge part of what I do with REBOL, 
so this is a great step.

It is just file!, no port! nor port's actions support yet.

Though, you can use Kaj's READ and WRITE functions to access files 
from Red already.

Got multiline strings LOADing working, but now I need to find a way 
to efficiently (if possible) output the right delimiters:

    probe load {
        {
    	Hello
    	World!
        }
    }

    == ["
            Hello
            World!
    "]

Wait, " strings are multilined too? How do you help the developer 
when multiline strings are an error?

By "help the developer" I mean trigger a syntax error, or log an 
error in your compiler, whatever. Errors help the developer.

You missed the point, Red's MOLD is not yet able to correctly handle 
multiline strings, I'm right now working on adding that support, 
I'm just wondering how to do it the best (same discussion than in 
SO chat with Ladislav).

OK, cool.

I'm not satisfied by the Rebol solution of scanning the string each 
time a MOLD is done, I wonder if there's any better solution.

That seems worth looking into.

Doc, in Red/System, I am currently forced to use the names READ and 
WRITE-string, because WRITE is taken for the syscall. Would you consider 
renaming that, to sys-write or os-write for example? I would really 
like to use READ and WRITE in Red/System like in Red

Not sure if you have time to do that before the release, but it would 
be good to avoid creating legacy code

I can do that, but I can't garantee you in the future that Red/System 
won't define its own READ / WRITE functions.

But for now it is unlikely, as I don't plan to add any file or network 
I/O support in Red/System itself. I do plan to use READ-IO and WRITE-IO 
though.

Doc, understood that file! is just the datatype right now. It's still 
an important step. :-)

Doc, certainly, I'll remove my READ and WRITE when you implement 
yours, like I did REDUCE and COMPOSE

LOAD doesn't understand empty strings:

red>> ""
*** Load Error: string ending delimiter not found!
== """
red>> print ""
*** Load Error: string ending delimiter not found!

The Red interpreter can execute the Fibonacci example now

I did some first measurements. The interpreter takes 1.5 times as 
long as R3

Good! :)

Given how much space for optimization we have, it's very good results.

Yes, that's actually an excellent result. We established before that 
the Red/System compiler is twice as slow as optimised C from GCC, 
so the interpreter could have been expected to be twice as slow as 
R3

So, is Red interpreter faster than R2 for fibonacci test?

It may reach that factor two once it does argument type checking, 
but a rough conclusion would have to be that the Red interpreter 
is pretty much at least as efficient as R3

Good question, I'll add a test. I would expect them to be about equally 
fast, as R3 is usually a third faster than R2

Just a third?? I thought it was twice as fast as R2 on average?

Nope

Diminishing returns. Almost half of REBOL's lifetime was spent on 
a one third speed increase

Good job, now BrianH will flood this group to prove you wrong! ;-)

Is there a way to hide this group from him (temporarily?)