Brock Pytlik wrote:
[email protected] wrote:
[snip]
server/catalog.py:
- line 291: Do we need EnvironmentError exception handling here?
Hmm, probably. What's the right behavior for a depot which is
missing one or more manifest files, and what's the right search
behavior? Without looking too deeply into the code, my guess is that
the right thing to do is to just drive on and serve the rest of the
information as search results, and log some kind of error. There's
no (simple/easy/reasonable) way for the indexing process, which is
(usually) a different process than the one serving requests, to tell
the other process to stop serving all search requests, and I'm not
even sure if that's the behavior we'd want anyway.
At least for now, logging an error is probably sufficient. My guess is
that if a depot is missing a manifest, the administrator is going to get
a lot of complaints from users. My concern was that we'd have a stack
trace and the indexing would never finish.
After some more thought, I'm declaring this issue not this bug. There
is the same amount of error handling now as there was before (ie,
none). I agree this is a problem, but the amount of plumbing needed to
make this work on the server side (and probably on the client side),
since we don't have a generic logging/error reporting module, is not
trivial. Rather than trying to shoehorn something into this fix, I
think it makes more sense to design this for carefully. Is that
reasonable?
Wish I'd seen __log in server/catalog.py before I sent that email. So,
there is logging available at least on the server side. Still, I'm going
to say it's not this bug and try to fix it in one shot in both the
server and client since it's not a regression.
Brock
Brock
[snip]
-j
_______________________________________________
pkg-discuss mailing list
[email protected]
http://mail.opensolaris.org/mailman/listinfo/pkg-discuss
_______________________________________________
pkg-discuss mailing list
[email protected]
http://mail.opensolaris.org/mailman/listinfo/pkg-discuss
