Hello Eric,
Thank you for the good suggestions, i'm working on implementing them,
I'm halfway out of the door right now (going to see "Toy story 2" :-)
I'll post it when done.
Nice to see that we share some opinions on this matter.
BTW.
has anyone else noticed what seems to be a bug in 'compose ?
(I havn't had time to follow the list closely for some time)
## system/version
== 2.2.0.1.1
## compose [([1])]
== [1]
shouldn't it return [[1]] ?
I've already reported it to [EMAIL PROTECTED]
Bye,
Thomas Jensen
-> off to the movies
On 03-Feb-00, [EMAIL PROTECTED] wrote:
>
> Hi Thomas,
>
> I like your idea on how to provide default values for optional
> arguments. I use arguments quite a bit in that way, and if FUNC
> did that automatically it would make the code a bit cleaner.
>
> The /tags refinement would make code involving calling functions with
> complicated refinements much easier to understand. All you would have to
> do is supply the one /tags refinement, and you wouldn't have to worry
> about which order the tags are in versus which order the arguments are in.
> The refinements would be right next to the values.
>
>
> A similar capability is already built into some functions, such as
> OPEN, where you can do:
>
> file: open/mode %test.txt [binary new]
>
> meaning the same as:
>
> file: open/binary/new %test.txt
>
> (This was from a message on the list from Jeff.)
>
> Unfortunately this /mode refinement doesn't seem to allow passing of
> arguments. It would be even more useful if you could do:
>
> file: open/mode %test.txt [new with #"^L"]
>
> to specify page-break as the line terminator.
>
> Just a couple more ideas. I looked at the source of the function
> you created:
>
> my-function: func [
> "My function"
> a [string!] "required argument"
> /b "A refinement" /c c-value [string!]
> {An optional argument with the default value "Hello"}
> /tags tag-list [block!]
> "A block of words to set, ie [size: 50]"][
> c: either c [c-value] ["Hello"]
> if tags [
> do bind (compose/deep tag-list) 'tags
> ]
> print [a c]
> ]
>
> The part
>
> c: either c [c-value] ["Hello"]
>
> Is not a good idea, since if somewhere else in the function you appended
> something to C, the string "Hello" would change. Also, there's a slightly
> more elegant way to do this check that I learned from a message from
> Gabriele. (Oops, I see he answered before me.) Maybe the resulting code
> should be something like:
>
> c: any [c-value copy "Hello"]
>
> But your function-making function would have to discriminate between
> datatypes that needed to be copied, and those that didn't.
>
>
> All this reminds me of another little idea I keep pushing:
>
>>> unset!: (type?)
> == unset!
>>> say-hello: func[a [string!] b [string! unset!]]
> [b: any [all [value? 'b b] "there!"] print [a b]]
>>> say-hello "Hey"
> Hey there!
>>> say-hello "Hey" "man!"
> Hey man!
>
> This kind of optional argument has the disadvantage that something
> that comes after might be taken as an unintended argument:
>
>>> say-hello "Hey" s2: square-root 2
> ** Script Error: say-hello expected b argument of type: string unset.
> ** Where: say-hello "Hey" s2: square-root 2
>
> so it's best in this case to always use the optional argument within
> a program, unless you wrap it in parens, or it's at the end of a block:
>
>>> (say-hello "Hey") s2: square-root 2
> Hey there!
> == 1.4142135623731
>>> if now/time < 1:00 [say-hello "Hey"] s2: square-root 2
> Hey there!
> == 1.4142135623731
>
> Nice talking to you,
> Eric
>
>
