World: r3wp

>> source d
d: func [][b/c]

Essentially:
	d: does bind [print c] context [c: "Cool"]

Yep.

>> d
Cool
>> source d
d: func [][print c]

But, like your version, if we can get the word, we can get its context:
>> probe do bind [self] second second :d
make object! [
    c: "Cool"
]

yep it can never be full proof in R2 as far as I know.

fool proof

ahhh wasn't sure, lol

actually "foolproof"

maybe fool-proof?

full proof
 pertains to strength of alcohol.

indeed

Well... we can unset 'self, so that the above method does not work:
>> do bind [unset 'self] second second :d
>> probe do bind [self] second second :d
** Script Error: self has no value
** Near: self

does that hide it completely?

I haven't tested that much at all, so I don't know if that would 
break anything as rebol goes along...

I'm not sure if there is another way in, quite possibly there is.

Even if it does it might be useful in controlled situations.

I know you can hide some things by simply declaring an object and 
afterwards setting the objects self to a different value

>> a/self: 3
== 3
>> a
>> probe a
make object! [
    b: 2
]
>> a/self
== 3

See my code above.

That still doesn't stop BIND? actually... Need to unset 'bind? as 
well :)

that could cause way to many problems I would assume.

If your code can reference a word, then code that treats your code 
as data can reference the word in R2.

If you want to sandbox R2, you have to sandbox the functions that 
can turn your code into data - that means the ordinals.

That will make the sandboxed code slow though, since it would mean 
PICK and the ordinals would be mezzanines.

R3 doesn't have that problem, since the ordinals are not used to 
turn code into data - another function is. Sandboxed code wouldn't 
need to have a reference to that function at all, or any functions 
that call it.

Working on locate.r;  saving a database after a scan of the library 
scripts can't be reloaded
autoextract.r has the following inside it:
    output: [{Self-extracting REBOL-compressed file
        REBOL [
            Title:  "Self-extracting compressed file"
            Date:  } now {
            File:  } mold infile {
            Author:  "Autoextract function by Bohdan Lechnowsky"
            Comment:  ^{
               Simply run this script and it will 
               decompress and save the file for you
            ^}
        ]

        if exists? } mold infile { [
            print ["} infile { already exists, please rename"
                           " existing file and run again."]
            halt
        ]
        write } mold infile { decompress 64#} mold file
    ]
    write outfile to-string reduce output


How do I get REBOL to keep the ^{ and ^}  across a save/all and load? 
  locate.r keeps a reference to all block! info for the tour sequence. 
 LOADing this (after a SAVE/ALL) causes an invalid string error. 
 Any hints?

Oh, and the code is passed thru the pretty-print parser (to build 
up the references) before the save/all

one of my pet peeves is that to-block craps out if it encounters 
an unknown datatype

why can't there be an option to turn illegal datatypes into strings 
...

I'm using to-block so that I can use block parsing of course

Yeah, me too ... always.   But R3 TRANSCODE can trap that now ... 
umm, I just don't get the whole LOAD/NEXT thingy in context of parsing 
over block! and paren!   Gabriele has posted links, just never dug 
in.

>> to-block "23 May, 2008"
** Syntax Error: Invalid word -- May,
** Near: (line 1) 23 May, 2008

Ok, now why is "," in particular not allowed in a word?

Has anyone ever detailed the voodoo of caret escapes in strings?


^^^{   does not return ^{  as I read it should.  It's ==ing as ^^{, 
but I'm not sure if this a post process of the == result display. 
 ??  Seems like voodoo.  And instead of exploring, I'd rather just 
read something this time.

Having nothing intelligent to add - to me the comma is a little voodoo 
too; I'll just ditto your sentiment.

which function allows me to traverse a series and remove at the same 
time?

remove-each

Or while?

cool

how's this 

>> date: "* 10 May, 2008"
== "* 10 May, 2008"
>> d: parse date none
== ["*" "10" "May" "2008"]
>> remove-each b d [ not any [ parse b alphas parse b digits ] ]
== ["10" "May" "2008"]
>> d: to-block form d
== [10 May 2008]

I'm trying to clean up OCR'd text prior to parsing

to end up only with integers and alphas ... is this bullet proof?

>> to-block "3a"
** Syntax Error: Invalid integer -- 3a

Not entirely...

it won't pass my parse rule

I parse digits and chars separately .. I'm not using an alphanumeric 
parse

ok, if that's the case, you should be fine.

Yeah I'm playing too ...

I'm sure the parse gurus can do this all in one parse rule!

I wanna junk! datatype ... parsed (made / loaded) up to next space 
during interpret.  Then we could read scripts modified by normal 
people.  Might be a lot of junk! but I'd rather write a junk! handler 
than try and trick REBOL.