> > I return into this already discussed item, because something
> > other is broken.
> >
> > Namely, ScanDeps contains "case-sensitive FS" heuristic also,
> > and it seem to fail in my case:
>
> This heuristic looks bogus, but what is it that actually
> fails for you?
Because heuristic looks bogus :)
> (AFAICT, finding "Tk.pm" when looking for the incorrectly deduced
> "tk.pm"
> has nothing to do with this and is in fact unavoidable when using
> a case-insensitive filesystem - perl itself would do likewise)
But this *is* avoidable with better heuristic:
- first it suspects for "tk.pm" because of string scanning "::tk::priv()"
- it searches tk.pm but file system gives "Tk.pm"
- at next step "tk.pm" ne "Tk.pm" (whereas lc("tk.pm") eq lc("Tk.pm")) so
wrong Tk.pm should be rejected here.
But if authors of the module invented "is_insensitive_fs" for something
different - then its okay with me.
But in my opinion differentiating between tk.pm and Tk.pm could be in scope of
is_insensitive_fs case...
Best regards,
Vadim.
