World: r3wp

Q1: I want to replace all 'none in a block with _different_ empty 
strings, what's the fastest way to do this?

(replace/all BLOCK none ""       replaces all 'none with the same 
empty string)

Q2: I have to blocks containing strings, and want to find out which 
of these strings differ (I need all differing positions), what do 
you think is the fastest way to achieve this?

Thanks in advance for all ideas!

PS. Do you, like me, feel that the replace way of doing things is 
questionable?

b: ["s1" "s2" "s3" "s1"]
parse b[any[to "s1" p: (p/1: copy "t0") skip]]
? b

Q2 again: Sorry, my axplanaition was a little unclear: 

I have a row from a database, and store away one block, and display 
the other for the user. The user may, or may not, change the data.
>> orig: ["Mr" "Petr" "Ustinov"]
>> data: ["Sir" "Peter" "Ustinov"]
>> magic-changed-func orig data
== [1 2]

orig: ["Mr" "Petr" "Ustinov"]
data: ["Sir" "Peter" "Ustinov"]
; same length !
out: copy[]
repeat i length? orig [ 
	if orig/:i <> data/:i [
		append out i
	]
] 
? out

Q1:use [x] [

Sorry

Q1:
use [x] [
    x: block
    while [x: find x none!] [change x copy ""]
]

Ingo, I have similar feelings too sometimes. I need a higher level 
function to do some things like that, but it's not there....

I often wanted reduce and compose to not make a copy for you, but 
work directly on the block. (Perhaps new functions "induce" and "impose" 
?)

induce should be fairly easy -- walk the block with 'for and 'poke 
back the value

blk: [a b c] 
format: [(1 + 2) (random 100) c]

impose: func [blk format][repeat n length? format [if paren? format/:n 
[poke blk n do format/:n]]]
impose blk format
;blk == [3 95 c]

For induce I think I would use do/next.

induce: func [blk format][change blk reduce format]  ; (except, as 
a native!, without creating a temporary block)
induce blk [1 + 2 random 100 'c]
; == [3 67 c]

Hi Voker, BrianH, thanks for your ideas ... they look mighty long 
compared to normal Rebol code ;-)

Sorry, I wanted to type Volker, of course!

; Q1: copy/deep will give you new strings:
block: [none none none]
new-block: copy/deep replace/all block 'none {}

; Q1 Or if you know you can reduce it you could do something obtuse 
like:
block: ["a" none "b" none none]
use [none][none: does [copy {}] bind block 'none]
reduce block

But if you do need to do an actual replace on the original block 
- see the other solutions :-)

Hi Brett, your first version looks pretty good ... now I'll have 
to dig out my little profiler, and see what's the fastest of all 
of these ;-)

I reckon the parse will win.

Brett's is the nicest looking though.

Brett, your obtuse way will only work if you are replacing 'none 
(as a word), but not none! values.

eg. reduce your input block:
	block: reduce ["a" none "b" none none]

Agreed Anton and perhaps that shows the obtuse method is not a great 
idea for this situation  - though perhaps an interesting one ;-)

In case anyone is interested... http://www.rebol.org/cgi-bin/cgiwrap/rebol/view-script.r?script=reboldiff.r

Are there any other instances where the a: copy [] vs. a: [] "problem" 
appears?

I'm have a bug where two arrays:

1. an array with data that can be manipulated

2. a similar one containing default data, which is used whenever 
I want to reset the first one


They apparently "stick" together with synchronized values. When I 
manipulate array 1, array 2 changes too. This would be the old COPY 
problem, but I use COPY everywhere, whenever I need to create array 
2.


However I do frequently PICK values from array 2 and POKE it in array 
1 at the same location to reset a specific location to a default 
value. Would that create a similar problem?

it depends on the kind of value you are picking/poking.

i guess you need copy/deep and poke ... copy pick ...

I pick and poke arrays

i mean, what is the result of PICK? i.e. what is inside the arrays?

if the arrays have multiple dimensions, you need copy/deep. And if 
there are objects inside, those objects are not copied, then you 
need explicit work (copying all objects in a loop using 'make)

Volker, the arrays are indeed both 2 dimensional and contain objects. 
I'll try a different init method.

gabriele, the arrays contain other arrays, hence PICK returns arrays 
of objects.

so, copy/deep, and, as volker suggests, you're likely to need to 
clone the objects too.

thanks, both. problem solved. :-)

a one liner I didn't see on rebol.com: Sum of all numbers in a block!:

do do replace/all mold [1 2 3 4 5] " " " + "

Clever!
But crucially dependent on the block being contained on one line:
 xx: {
 do do replace/all mold [1 2 3
  4 5] " " " + "
 }
 do xx

But then you did call it a one-liner :-)

do trim/lines xx could fix that

Is there a function which cuts the suffix, so i can go from %file.txt 
to %file.html ? I  remember darkly there is one now.

I think there is one which splits a file up in an object with path, 
filename and extension. can't remember which though....

thats decode-url, good thought, but not in my case.

what about cutting at last occurance of "dot"?

longer :) but thats what i do now.

>> prefix: func [filename][copy/part file find/last file "."]
>> prefix file
== %movie.cd1.divx-rel6

where - >> file: %movie.cd1.divx-rel6.avi

maybe 'prefix? would be good addition to Core, as we have 'suffix? 
already? :-)

thats cool, better than mine :) thanks.