Re: [TCLCORE] catch and "return -code" usage

[Neil M. <ne...@cs...>]

[break] and [return -code] are two different things. Example:

catch {break}  ;# -> 3 (break)
catch {return -code break} ;# -> 2 (return)
proc foo {} { break }
catch {foo} ;# -> 1 (error - invoked break outside loop)
proc bar {} { return -code break }
catch {bar} ;# -> 3 (break)

The return -code stuff causes a TCL_RETURN return code, which is then=20
converted to a TCL_BREAK return code after one stack level has unwound.=20=

To get them to be equivalent, you can use (in 8.5) [return -code break=20=

-level 0]. The reason for the difference (without the -level stuff) is=20=

so that stuff like this works:

proc MyBreak {} {
     return -code break
}

This will work just like the built-in [break] command.

On 24 Jul 2004, at 17:00, Martin Lemburg wrote:
>  Hello Lars,
>
>  you are right, the TIP 90 is what I ment, but it is very difficult to=20=

> understand this document..
>
>  In my eyes, it is not a lack of functionality, but more or less a=20
> bug, that catching e.g. "break" returns the right value, but catching=20=

> "return -code break" always returns not the matching value.
>
>  Why a TIP and not a simple bugfix to let the catch command return the=20=

> right value?
>
>  The TIP 90 is about extending the "return" and "catch" functionality,=20=

> right?
>  My question is about the normal, regular and expected behaviour of=20
> the combination of "catch" and "return -code ...".
>
>  Ok - in v8.5 the wanted functionality exists, so I have only to wait=20=

> and then to switch to the this version.
>
>  No problem for my private projects, a big problem for the company=20
> projects!
>
>  Thanks and best regards,
>
>  Martin Lemburg
>
>
>
>  Lars Hellstr=F6m wrote:
> At 01.29 +0200 04-07-24, Martin Lemburg wrote:
> [stuff deleted]
>
> ... is not conform to the man pages of the return and the catch=20
> command,
> because the catch command should return the values of the exceptional
> return codes. And in the return man page is to be read, that the -code
> option allows to create exceptional return codes.
>
> So - what is going wrong? Should I file this as a bug on=20
> sourceforge.net,
> or is this already known and documented in the bug database, or=20
> corrected
> in v8.4.6+?
>
> And - I see the differences from "return" and "return -code return",=20=

> but
> how should I differentiate via "catch" these two cases?
>
> Take a look at
>   TIP #90: Enable [return -code] in Control Structure Procs
>   http://purl.org/tcl/tip/90
> The descriptions there should answer your questions.
>
> Lars Hellstr=F6m
>
>
>
>
>
> =20=