World: r3wp

Robert - I think not. Henrik named few fundamental changes Carl is 
after. You surely remember e.g. frames concept, which was later removed 
for e.g. IIRC, things like - better resizing, changes to layout engine, 
layers, etc. are planned. What is interesting is, that it would take 
imo max 1 week to implement them. So - dunno when Carl is back on 
GUI. As for me, I would not like him to jump from topic to topic, 
which happened today, as Carl is moving to website topic for few 
days. BUT! - I think he just needs break from heavy coding ...

Jerry, from observation, Carl fixes bugs related to specific topics, 
so at some point, I'm sure he'll revisit ports as long as people 
put clear and simple reports in Curecode.

it should read: "specific topics in one go"

So I guess this bug won't be fixed soon.

 Also not easy to know. I've before put reports in that were fixed 
 on the same day. (But it definitely won't be fixed, if it's not in 
 curecode).

Henrik - it's there, actually two tickets, probably the same #1369 
and #1397

good

R3 ports don't do auto-lookup? Hmm, maybe it is logical, on low-level. 
I tried Oldes' example in CC ticket, and something like open tcp://localhost:8080 
does not work here ...

those examples are bad and difficult to use. I should probably port 
my short "multiserver" script to R3 and give it a try ...

so try another port! that's the only reason why  open tcp://localhost:8080 
 should not work

And the example is the simplest I have been able to produce... if 
you have better, just write it to comments.

And I agree with Jerry, that it's quite urgent bug. It's hard to 
start writing some network schemes when you can simply test, that 
there is something wrong inside. If someone is able to "beep" Carl 
a little bit, please do it.

I think that OSX host kit can wait a little bit.

beep
 Carl: Mark the bug urgent.

Another IP related bug is affecting the amount of transferred data, 
when you try a few megabytes you can find the error just after 4300000 
bytes or 215 chunks, I've reproduced this a few times.

Unfortunately this I can't write to curecode because my login is 
unactivated still, where to activate it?

you should have gotten an email from the curecode server

if not, then poke DocKimbel in the Curecode group.

Anyvay when you play with demo IP applications you can find that 
rebol doesn't clean memory the consumed memory grows slowly but after 
few hours/days you could be short of memory IMO. Not good for serious 
(server) application.

That sounds like a "good" bug. :-)

it is still possible that rebol does some garbage collection in some 
period/treshold but I'm not sure

there are some stats that you can output to tell whether GC has been 
performed.

stats/profile, I believe.

OK thanks Henrik, I'd try it.

PeterWood: I am missing the formulations from the doc articles you 
mentioned, that are in conflict with your result

Some example of using to-get-word, to-set-word, to-lit-word? what 
it is really for? especially in conjunction with extensions.

Some description would be helpful

I use them when building blocks, where 'reduce will remove set-words, 
get-words and reduces lit-words to workds.

So the resulting block seems as original one after reduce? short 
example pls. TIA

>> reduce [a: 2]
== [2] ; not what we want

>> reduce [to-set-word 'a 2] 
== [a: 2] ; what we want


Useful when building REBOL code or objects bit by bit. In R3 with 
extensible objects, you don't have to do that, though.

And given the ability to create REBOL code, you should be able to 
generate any datatype with a to-* function.

it's usefull when you are generating code from data

well, the example

    reduce [to-set-word 'a 2]

is not R3-ish, in R3 it could have been:

    reduce [quote a: 2]

oh, didn't see that one before.

yes, Quote is a new function. A mezzanine, actually

See the source, Henrik

doh :-)

Why to use reduce when you must defend your words by aditional armour 
:o

as follows: reduce [quote a: 1 + 1]

, i.e. in situations, where you actually do want to compute some 
parts of the block

nevertheless, I do not think, that Reduce (or Compose) is ideal, 
I proposed an alternative

in Reduce you have to "mark" what should be kept, while I prefer 
to mark what should be computed

Meaning extensions this should be used for some lexical operations 
or code trimming, I've think abour extensions in pure functional 
way before

(which is the case of Compose, except for the fact, that it does 
not handle parens good enough for me)

It is true that there are always new features that it takes a little 
time to understand the usefulness.
I'm caught regularly.

Even if I had noticed the existence of QUOTE, I saw no interest until 
the Ladislav's example.

glad to advertise this useful new feature

I suppose 'build is kind of an opposite to 'reduce.

yes, or, stated differently, it is a kind of improved Compose

Perhaps a shortcul should be defined. Because I feel that it will 
be often used as an alternative to COMPOSE.
Is this name QUOTE not a bit too long ?

(i.e. my idea, how Compose should have worked)

it has a few gotchas though. can they be removed?

*shortcut