Re: Weird result on using define-class twice

[Thomas Wawrzinek]

Hi!

Thien-Thi Nguyen <address@hidden> wrote:

 > actually, there are no <bar> objects at this point, so redefining the
 > class is no big deal.  however...
 > 
 >   (define baz (make <bar> #:b-value "BAZ"))
 >
 > i would think if an error is to be signalled, it would be here, in the
 > form of a "no such accessor" error.

Sounds reasonable. It could at least put you on the right track.

 > would it be correct to surmise that such a case can be detected
 > statically?  if so, probably we need to write "guile-tools lint".

I suppose a statical analysis would be possible. 

Would this involve analyzing the whole program, or only individual source
files (consider load)?

ISTR there was an agreement that name conflicts resulting from different
modules exporting the same "names" should raise a warning, so at least 
this case would be covered once the warning mechanism is 
implemented ...

Regards,

                        Thomas