World: r3wp

I copied it to death until I figured out it was in loop.

Maybe not a bug just a need to use copy/deep in loop.

Steeve, I see what you were saying.  I went ahead and fixed that 
by changing a piece of code so that the d block wasn't modified also.

It's not a bg in loop, you are reusing the inner block [fname] where 
you shouldn't.

>> d: []
== []
>> loop 2 [append d [1 [fname]]]
== [1 [fname] 1 [fname]]
>> same? second d fourth d
== true

Yeah we already determined  that.  Thanks Brian.

Cool :) I couldn't tell based on the messages above.

Yeah Steeve keyed in on the problem.

Is there a way to modify an object (adding and removing components) 
at runtime without rebuild it from scratch ?

I remember an old thread in the mailing list abut this argument, 
but I cannot find it..

not it's impossible in R2 but doable in R3

in R3 do:   append my-object new-var

wrong, in R3: append my-object [new-var: none]

R3 is going to be my nightmare :-)

when i need to simulate extendable objects in R2, i use blocks in 
replacement.

I'm trying to find a workaroud using first object , second object 
... but it doesn't seems to work

it can't

I'm writing a function that  replace every set-word in a block of 
code with a compoment of an object, so I have to build the object 
in one single pass

(so every words become local for default)

are there inner blocks which contain set-words too or not ?

yes there are

so you can't use the construct function on the primary bock...

except if the local vars are all set in the primary one too

the code is an handler so it could be everything

y

I'm looking in lfulc code by ladislav, the parse rule to catch the 
set-words is quite simple

lfunc

you could write your handler in a context directly, it's the most 
simple thing to do

it's in a context

but the words used are still global, no ?

not those which are scoped directly by the context

ctx: context [
a: 1
block-code: [a: a + 1]
]

a is defined local to the context ctx

I'm not sure if this can apply to my app. It's a web server so I 
have to mantain client values between session

the context is unset after single request

and ??? it's not related with your previous request

If I change the code as:

ctx: context [
saveobj/a: 1
block-code: [saveobj/a: saveobj/a + 1]
] 

I can retain the counter between web session

you lost your context, because you don't keep him  in memory (i don't 
know how to answer to this problem)

saveobj is not in your context

ctx: context [
localobj: saveobj
localobj/a: 1
block-code: [localobj/a: localobj/a + 1]
]

this should work...

yes ! saveobj is global, so I can use it for saving values that are 
in my context . In my context values will be lost. I'm writing a 
function to change every var in my handler like "var" to "saveobj/var"

i don't see y

i don't see your point... ctx is global too... you shoul'd not lose 
it

Yes, I'm not too clear .... I unset the handler block every request 
I get

so If I need to keep the values of the hadler I need to store them 
in a global ojb

i think you have several problems not not necessarily linked, you 
should have a look on rebol.org, there are many web servers examples

sorry...

Wow another webserver.  Does REBOL have the record in webservers?

thanks ! only talking about this cleared my mind, I'm thinking about 
keeping the handler in memory

That is really cool Davide.

surely ;-)

thanks paul, this is not a standard web server... it's a comet web 
server (where the browser use an endless connection)