--- Benoit Boissinot <bboissin@xxxxxxxxx> wrote:
> On 10/29/05, TK Soh <teekaysoh@xxxxxxxxx> wrote:
> > --- Benoit Boissinot <bboissin@xxxxxxxxx> wrote:
> > > Did you actually hit that bug ? because the doc says that is allowed
> > > to pass a filename instead of a list, and it works for me.
> >
> > Not actually a bug in ConfigParser, but without it the error message would
> show
> > both /etc/mercurial/hgrc and $HOME/.hgrc files, when either of them fails
> to
> > parse.
> >
> ok, then maybe it can be better to fix all caller and have a unique
> interface (doing the "for f in filename" in __init__).
That'd make little difference, IMO.
> The recent traceback fixes raise a question: when should hg fail and
> when should it just skip an invalid input ?
> For example, it skips invalid extension, but it fails with invalid .hgignore.
>
> Do we need to be more consistent ?
Offhand, I'd say making Hg fail is the better choice. Ignoring the bad hgrc may
create more problems than it solves. Besides, that way it would certainly get
the user's attention to report/fix the broken hgrc file ASAP.
__________________________________
Yahoo! FareChase: Search multiple travel sites in one click.
http://farechase.yahoo.com
|
