Re: PEP proposal: Per-Module Import Path
Eric Snow <[email protected]>
| Newsgroups | gmane.comp.python.import |
|---|---|
| Message-ID | <CALFfu7BuCgM1b-AAo8NKaaGD8vWSTA_=c1Sgq4WRSh85vZ3Crg@mail.gmail.com> |
On Wed, Jul 31, 2013 at 6:29 AM, Brett Cannon <[email protected]> wrote: > On Wed, Jul 31, 2013 at 2:41 AM, Eric Snow <[email protected]>wrote: > >> I'll make it more clear in the PEP that refpath resolution is >> depth-first--a consequence of doing normal loader lookup. This means that >> in the face of a cycle the normal package/module/ns package handling >> happens rather than acting like there was an empty .ref file (but only if >> no other path entries in the .ref file pan out first). Would it be better >> to treat this case the same as an empty .ref file? >> > > I would argue a cycle is an error and you should raise ImportError. > I'll make a point of this. An ImportError makes sense. Either way I'll provide some rationale. -eric _______________________________________________ Import-SIG mailing list [email protected] http://mail.python.org/mailman/listinfo/import-sig