World: r3wp

Is this prefered result?:
>> sort [748x430 68x206 748x194 305x147]
== [305x147 748x194 68x206 748x430]

I take it back.. REBOL cannot have any idea what I want to do, so 
better to use compare function, like:
sort-pairs: func[block-of-pairs][
	sort/compare block-of-pairs func[a b][
		either a/x < b/x [
			1
		][
			either a/x = b/x [
				either a/y < b/y [
					1
				][
					either a/y > b/y [-1][0]
				]
			][ -1 ]
		]
	]
]

sort-pairs [748x430 68x206 748x194 305x147]
;== [748x430 748x194 305x147 68x206]

What object issues?

XML to objects from SUnday

get-obj-value: func[o [object!] p [path! word!]][attempt[do append 
to-path 'o p]]

Not optimized for speed though...

hmm... clever

So why am I walking the path?  lol

Oldes, exactly the use case to use case... 

case [
	a/x < b/x [-1]
	a/x > b/x [1]
	a/y < b/y [-1]
	a/y > b/y [1]
	'equal 0
]

(except I reversed some stuffs...)

yes, that's nice example where to use case:)

Steeve, that's sweet. "Case" doesn't show up in the dictionary and 
I had no clue we had such a thing. Very cool.

not in the dictionnary ? strange...

http://www.rebol.com/docs/dictionary.html

I use it a lot. CASE/ALL is also useful.

CASE is in the R3 dictionary:
   http://www.rebol.com/r3/docs/functions.html
But you are right -- it should be in the R2 dictionary too.

anyone know how to submit args when using 


r2.exe --do "probe what-dir probe system/script/args" --args  "r3.exe"

the above prints none

The help says this:
    -- args          Provide args without a script

There's a space between the -- and the word args, so I tried this 
and it worked:

   rebol --do '? system/script/args q' -- 'arguments here'

The help should probably say something like:
   -- <args>
to make it more obvious

doh... ok, I get it.

thanks.

Isn't this a rather erroneous message ?

>> now - 365
== 27-Oct-2009/23:03:45+13:00
>> now - 365.25
** Script Error: Cannot use subtract on date! value
** Near: now - 365.25

Yes.. Better might be "Cannot subtract decimal! from date! value."

email? [name-:-domain-:-com]
== true
>> email? [name-:-domain]..com
== true
>> email? [name-:-domain]...com
== true

Is this a bug?

The email parser is not very good.

You can convert anything to email.

I was going crazy trying to send to a long list getting errors...finally 
found the faulty address...how do you validate email then?

You should write your own parse rule, it should be failry simple. 
Probably there's already something on http://www.rebol.org

Amacleod - there was a discussion on ML or elsewhere, about how useless 
email dtype is, if it can't work as email RFC suggests. I was told, 
that I should not mistake datatype, with complicated parser for possible 
correct emails. I still insist - the datatype is useless that way. 
I found some grammar, I even posted it back at that time, but I think 
that someone at RT was simply too lazy to implement it :-)

the email datatype is not useful for different reasons: it can't 
be serialized properly.

either write your own parser ... or use my smtp challenge :)

What would you expect here?

>> ?? test3: now
test3:
== 28-Oct-2010/17:28:36+13:00
>> ?? test3

none

or unset

why?

?? prints test 3 and now is returned evaluated after so its printed 
by the console

so where is the output of ?? test3:

ahh... on the first line

>> ?? test3: now
test3:   <<------ here
== 28-Oct-2010/17:28:36+13:00

its a common error I make ever so often in code.

because ?? is a lit-word argument.

(uses)

it grabs 'test3: directly, so it never gets evaluated (so it doesn't 
assign the following value)

this is true of all lit-word arguments and is probably why we don't 
use them often, they brake the visible chain of command.

I should just inline 'probe

yep.. usually a better alternative... and put a prin "test3: " on 
the previous line.

(previous line is for scripts, obviously makes no sense for the command 
line hehehe)

I generally only use ? and ?? for words when I want to know the type 
as well, and because of that very issue I started using this:
probe: func [D][
    print join join type? D "!  " mold :D
    D
]

It's in my primary include file.  Helps with 'none versus none!, 
etc...

the only thing is the loss of the variable name, but using lit-word 
argument breaks the entire reason for using this, so... eh.  It'd 
be very nice to have some sort of in-between

I have redefine my probe so it uses mold/all instead... it also only 
dumps the first 500 chars when I try to probe something which is 
way to big (like a nested set of faces)

what's the word to remove one series from another?

someop [ 1 1 2 3 4 ] [ 1 ] 
== [ 1 2 3 4 ]