World: r3wp

>> str: "this is just a test of replace-all"
== "this is just a test of replace-all"
>> replace-all str #"t" #"d"
== "dhis is jusd a desd of replace-all"

The case did cut it down a bit on the evals:
>> replace-all b 1 2
== [[2] [[[2]]] [2]]
>> stats/evals
== [209 100 42]

I would say that is one heck of a useful function

I've been using timblk to profile, not stats. How useful have you 
found stats to be?

its pretty useful.  Btiffin mentioned timblk also. I will have to 
look for it sometime.

replace-all: func [
    "Replaces all occurences of old value with new value"
    series "Series containing values to replace"
    oldval "Value to be replaced"
    newval "New value to replace old value"
    /local subf
    ][
        subf: func [sd][
            while [not tail? sd ][  
                case [
                    equal? first sd :oldval [poke sd 1 :newval]
                    series? first sd [subf first sd]
                ]
                sd: next sd
            ]
        ]
        subf series
        series
    ]

same function just commented a bit and changed sub-ic to just subf 
meaning subfunction

catch throw-on-error?

Add the [series!] type check to the series parameter of the outer 
function and you'll be set.

oh yeah

catch throw-on-error
 What do you mean?

Should it throw errors to a catch so that it doesn't expose the lower 
code?

not sure how that would behave in this case

You aren't doing anything in the inner code that could generate an 
error, so no need.

If were sure that all the values that it traverses are comptible 
then I agree.  In fact I don't think we need one until it becomes 
known as a problem.

The only problem would be if your blocks have cyclic references and 
you get stack overflow. I'm not sure you can catch that.

replace-all: func [
    "Replaces all occurences of old value with new value"
    series [series!] "Series containing values to replace"
    oldval [any-type!] "Value to be replaced"
    newval [any-type!] "New value to replace old value"
    /local subf
    ][
        subf: func [sd][
            while [not tail? sd ][  
                case [
                    equal? first sd :oldval [poke sd 1 :newval]
                    series? first sd [subf first sd]
                ]
                sd: next sd
            ]
        ]
        subf series
        series
    ]

Don't put [any-type!]. The only difference between that and no type 
spec at all is that your function would be able to accept unset! 
values, and that would require other changes to your code to work 
properly.

here this is better to read:

ahhh good point.

replace-all: func [
    "Replaces all occurences of old value with new value"
    series [series!] "Series containing values to replace"
    old-value  "Value to be replaced"
    new-value  "New value to replace old value"
    /local subf
    ][
        subf: func [sd][
            while [not tail? sd ][  
                case [

                    equal? first sd :old-value [poke sd 1 :new-value]
                    series? first sd [subf first sd]
                ]
                sd: next sd
            ]
        ]
        subf series
        series
    ]

I changed newval to new-value and oldval to old-value

That is the REBOL way :)

hehe - looks good.

In general, I find that it is a good idea to let unset! values be 
erroneous, rather than changing your code to accept them. That makes 
it easier to make a clear distinction between non-values that are 
erroneous (unset!) and non-values that may not be erroneous (none!).

good tip

Is it a good practice to unset series within functions if the series 
is an argument to the function?  Not sure how the garbage handler 
in REBOL works or if it sees it as garbage until the next function 
call.

In other words if i pass a massive series of data as an argument 
to the function should I at the end of the function unset that argument?

I wish Carl would give us pointers on when to use unset function 
and how to optimize our use of the memory space and other performance 
tips.

That was something I brought up during R3 development. R3 series 
don't retain references between calls, R2 series do. In general this 
is not a problem, but if the extra references are becoming a memory 
leak, change this:
    series
to this:
    also series series: none

what is also?

You almost never need to use the UNSET function; setting to none 
is usually sufficient.

interesting.

ALSO is one of the R3 backports in 2.7.6. Here:
also: func [
    {Returns the first value, but also evaluates the second.}
    value1 [any-type!]
    value2 [any-type!]
][
    get/any 'value1
]

ok I remember some discussion about that.

This is one of those cases where unset! wouldn't be an error, so 
the code handles it.

It would be any problem? wouldn't the get/any cause a problem if 
it encountered the unset!?

That's what the /any is for.

indeed

don't know that include unset!

don't = didn't

RobertS, have you tried initialization functions?

I think your INITIAL and INITIALLY functions could be combined though.

initially: func [[throw] tag [word!] code [block!] /local tags] [
    tags: []
    unless find tags tag [
        insert tail tags tag
        do code
    ]
]

I suppose a catch attribute would be appropriate to add to the function 
too.

Or you could eliminate the tag:
initially: func [[catch throw] code [block!] /local done] [
    done: []
    unless find done code [
        insert tail done code
        do code
    ]
]

Sorry, not insert, insert/only.

It relies on the FIND finding blocks based on whether they are the 
same, not equal. That means that the reference to the code block 
that is passed to INITIALLY can itself be used as a tag.

Final version:
initially: func [[catch throw] code [block!] /local done] [
    done: []
    unless find done code [
        insert/only tail done code
        do code
    ]
]

Rambo 3115 submitted for desire to include the replace-all function.