Hi, Romano,
Romano Paolo Tenca wrote:
>
> You can do also:
>
> name: "Guido"
> print favoriteLanguage/:name
>
> The path evaluator make an implict select on block and hash.
> But the problem here is that if a language called "Guido" appears
> before the user "Guido" then fails...
>
Given
>> name: "Perl"
== "Perl"
>> favoriteLanguage/:name
== "Carl"
I think I'd say "gives an unexpected result" rather than "fails",
since there's not explicit error involved.
>
> A fetch based on this:
>
> fetch: func [k [string!]] [data/:k ]
>
> but if k is not found fetch will fire an error (Invalid path value).
>
> It is possible to do also a change, if the index is a word (it
> seems to me a strange thing - what do you think?):
>
> data: [guido "Rebol"]
> change: func [k [word!] value] [k: to-set-word k data/:k :value]
> change 'guido "What else?"
>
There are a number of "strange things" in the use of paths; using
classSize:
[ "freshman" 166
"sophomore" 141
"junior" 122
"senior" 131
]
we can say
>> select classSize "junior"
== 122
>> select classSize "graduate"
== none
compared with
>> x: "junior" classSize/:x
== 122
>> x: "graduate" classSize/:x
** Script Error: Invalid path value: graduate
** Where: halt-view
** Near: classSize/:x
But when using
classSize:
[ 9 166
10 141
11 122
12 131
]
we find
>> select classSize 7
== none
>> select classSize 9
== 166
compared with
>> x: 7 classSize/:x
== 12
>> x: 9 classSize/:x
== none
and
>> pick classSize 7
== 12
>> pick classSize 9
== none
I don't understand the purpose of the inconsistencies:
1) when to generate an error vs. returning NONE to indicate
data-not-found, and
2) the fact that /: performs access-by-position or access-by-key
depending on the type of the second word's value.
Multiple ways to do not-quite-the-same-things seem to be an
invitation to confusion.
-jn-
--
To unsubscribe from this list, please send an email to
[EMAIL PROTECTED] with "unsubscribe" in the
subject, without the quotes.
