World: r3wp

Use FORALL for the R3 version of GATHER - FORALL is faster than FOREACH 
in R3.

ASSEMBLE might be a nice word to reserve.

Could be disturbing for beginners, a same thing can be writed  using 
so many various ways with the tons of mezz we have now.

Yeah. For an assembler :) We want reserve COMPILE too :D

Steeve, you have a point. For a couple of years, I would write mezzanines 
for things that already existed in R2, but hadn't noticed.

ahah, shame on you ;-)

I'm converting GATHER to mezzanine/library code standards now. We'll 
probably put all of these is an advanced series manipulation library 
module.

types-of is ugly Brian :)

You have no idea. TYPES-OF was the only reflector that had no corresponding 
hack in R2. I had to do it from scratch.

Mezzanine-quality version of Henrik's GATHER, R3 version:

gather: func [
	"Get the values of a given field from all objects in a block."
	block [block!]
	word [word!]

 /into "Insert into a buffer instead (returns position after insert)"
	output [series!] "The buffer series (modified)"
][
	unless output [output: make block length? block]
	forall block [
		all [
			object? block/1
			in block/1 word
			output: insert/only output select block/1 word
		]
	]
	either into [output] [head output]
]

Has the spec for make function! been changed on purpose in R3? Is 
for what reason?

Is => if

>> a: make function! [spec] [body]

** Script error: cannot MAKE/TO function! from: function!

** Where: make

** Near: make function! [spec] [body]

>> my-func: make function! [[b] [print b]
]
== make function! [[b][print b]]

Yes. Reason: To simplify and speed up MAKE, and REBOL, and to make 
code generation easier.

Use the wrappers if you want to be R2 compatible.

Brian, are you sure that (output: insert/only output ..) is faster 
than (append/only output) in R3 ?

No, it's not, but not by much and it's necessary for the /into option.

ah yes, i forget

Strangely enough the REBOL overhead of the /into option (one UNLESS, 
one EITHER, and n assignments) is dwarfed by the overhead saved if 
the /into option is implemented pervasively.

Plus, I can copy-paste most of the code :)

and you say that foreach is slower than forall in this case ? (overhead 
with 3 path evaluations block/1 , no ?)

BrianH, nice GATHER function. :-)

FOREACH does a BIND/copy of its code block, but you might have a 
point in this case because the 3 paths are N*3.

Final submitted R3 version:

gather: func [

 "Get the values of a given field from all objects in a block that 
 have it."
	block [block!] "A block which may contain objects"
	word [word!] "The field to look for"

 /into "Insert into a buffer instead (returns position after insert)"
	output [series!] "The buffer series (modified)"
][
	unless output [output: make block length? block]
	foreach item block [all [
		object? :item
		in item word
		output: insert/only output select item word
	]]
	either into [output] [head output]
]

So FOREACH is better, anyway?

not anyway

Argh, I need to read better. You wrote this already.

It seems like speed is more important than memory in this function.

Henrik, if you can write up a CureCode wish for this function I can 
mark it as pending.

For GATHER? Should I just copy your code and credit it to myself? 
:-)

yes, we all do that :)

Write up a descriptive request, explaining why you need it, or reference 
your R3 chat message number.

Sounds bureaucratic, but OK :-)

CureCode is documentation of the new functions. We need the bureaucracy 
for the users' sake.

If we don't have a record of why the function was added we won't 
know where to puut it when we modularize R3.

Posted.

Henrik, don't forget to add: "Approved by the politburo"

:-)

And 10 lines for signatures.

That's my job (read the ticket comments :)

Curecode needs a trigger for rebdev links.

It will get one, but for now I say something like "R3 chat 1436".

That is the /into option message, so I remember it :)

Here's the R2 version of GATHER:

gather: func [

 "Get the values of a given field from all objects in a block that 
 have it."
	block [block!] "A block which may contain objects"
	word [word!] "The field to look for"

 /into "Insert into a buffer instead (returns position after insert)"
	output [series!] "The buffer series (modified)"
][
	unless output [output: make block length? block]
	foreach item block [all [
		object? get/any 'item
		in item word
		output: insert/only output get/any in item word
	]]
	either into [output] [head output]
]

Do you want a /deep option?

That will require switching to PARSE for speed.

I don't think it's worth it.

for me it makes the most sense to gather values from a particular 
level.

We can always add another CureCode wish for /deep if we need it. 
The current function is fast, so that's nice.

I marked the ticket as pending, with the particular submitted files 
that contain it.