Tcl Source Code

[tailcall next] complains spuriously that 'next may only be called from inside a method' even when it is.  See attached.

I think I have another problem related to this, where [tailcall next] is complaining spuriously 'wrong # args: should be "next"', but will see how this one pans out.

allow_comments - 1

[tailcall foo] is like [return [uplevel foo]] apart from some quoting issues and the context in which [foo] is resolved. (But there's only one [next], so the resolution will be consistent.) Thus, [tailcall next] only works at all when the _caller_ of the object can do a [next], which means that it needs to be done inside a method called from inside a method. I'm guessing that that's something you don't need. :-) It's complaining about the wrong # args because that's what the caller's next method implementation wants (I guess).

So... it works, but not in the way you would reasonably want. (Remember, [next] effectively does an [uplevel] inside itself so methods don't need to know about potential subclasses overriding them. It also _requires_ that the caller be an object method body context, since it uses that to know what is going on.) To do anything else would require teaching [tailcall] to know specially about [next] and vice versa, which would be rather disgustingly nasty.

I guess, by the time the [next] is called, the [tailcall] has already taken effect, and somehow the context is no longer that of an object.  Shouldn't it be, though?

File Added - 434664: tailcall.tcl

File Deleted - 434661:

File Deleted - 434663:

File Added - 434663: tailcall.tcl

File Added - 434661: tailcall.tcl