Re: expr functions

> Le 14/08/2012 16:30, Brian Griffin a écrit :
>> This question just occurred to me:  why is (should) expr functions (be)
>> restricted to ::tcl::mathfunc:: ?  Why could it not use the same scoping
>> rules Tcl uses for commands?
>>
>> I got to this point after noticing, first, a lot of occurrences of "if
>> {[expr {$a + $b}] > 3} ...", i.e. unnecessary calls to [expr] within
>> [expr].
>> After cleaning these up, I noticed unclean-able cases of "if {[lindex
>> $list [expr {$a+$b}]] ne "foo"} ..." where the recursive call to [expr]
>> is unavoidable.  This is solved by creating an alias to lindex in
>> ::tcl::mathfunc::, so now it can be written:  "if {lindex($list,$a+$b)
>> ne "foo"} ..."
>> Then I started noticing "if {[myfunc [expr {$a+$b}]] == 3} ...".  At
>> this point it starts getting ridiculous trying to simplify and make
>> these expressions more reasonable to read, except if [expr] would just
>> use the normal search rules to find functions...
>>
>> Thoughts?
>
> A quick & dirty proof of concept with unknown handlers:
>
>
>     proc mathproc {cmd args} {
>         if {[namespace qualifiers $cmd] eq "tcl::mathfunc"} {
>             return [uplevel [list [namespace tail $cmd]] $args]
>         }
>         uplevel [list ::unknown $cmd] $args
>     }
>     namespace unknown mathproc
>
>     set l [expr {list(1,2,3)}]
>     => 1 2 3
>
>     expr {lindex($l,2)}
>     => 3
>
>     expr {expr(join($l,"+"))}
>     => 6
>
>
> I love Tcl !!!