Roland Mainz wrote:
> Dan Mick wrote:
>>>> We use fnmatch() from the fnmatch module. I don't know what that uses
>>>> underneath. But in any case, if it *did* use libc::regex(), I certainly
>>>> wouldn't want to switch it underneath since that would change the
>>>> semantics of fnmatch() for all Python programs.
>>> Erm, aren't these patterm platform-specific anyway (e.g. shell pattern
>>> for Unix/Linux differ from Windows shell pattern etc.) ?
>> Erm, why would that matter?
>
> See comment above ("... would change the semantics of fnmatch() for all
> Python programs ..."). If the pattern matching is platform-specific
> anyway and we change the behaviour for all python applications on
> Solaris... why would this be a problem (note that "~(" is an invalid
> pattern anyway, e.g. this extension cannot break existing applications)
> ?
if we change RE behavior for all Python programs, whether it's only on
Solaris or not, that changes all Python programs.
If the change is completely backward-compatible (although I can't see how
it can be, exactly), then why would it matter if it were Solaris or every
platform?
i.e: platform-specific is a red herring, but changing RE behavior is not
acceptable, I would think.
Why is ~( "invalid"?
_______________________________________________
opensolaris-code mailing list
[email protected]
http://mail.opensolaris.org/mailman/listinfo/opensolaris-code
