[BUG] testfetch.py doesn't handle EAGAIN failures

[John A Meinel]

Martin Pool wrote:
> On 21/10/05, John A Meinel <john at arbash-meinel.com> wrote:
>
>>On my Mac box (and on Windows), I frequently got EAGAIN failures. Which
>>are getting written into the log, such that testfetch.py fails to see
>>what it is looking for.
>>
>>I'm guessing the best thing is to just not log EAGAIN. But if that isn't
>>possible, then I guess just filtering out the EAGAIN messages before
>>testing to make sure things are only requested one time.
>
>
> I don't think I've ever seen that on Linux.  (I thought that Python
> handled the EAGAIN for you, but obviously not.)  Could you post a bit
> of the log?

Well, actually, I've never gotten EAGAIN on linux, but I have gotten it
on both Windows and Mac. I was the one who added the code to handle it.
in bzrlib/selftest/HTTPTestUtil.py
I had to override the handle_one_request() function, so that it would
trap the EAGAIN value and actually try again.

If you wanted, we could just not log the fact that EAGAIN happened (I
would be okay with that), and then I think the weblog would be accurate
again.

>
> For concrete bugs I'd like to suggest people file them into "Malone",
> our bugtracker, here:
>
> https://launchpad.net/products/bzr/+bugs
>
> But if this is just "something to watch out for" it may not need an entry yet.

It's a bug in the test suite, not in bzr itself. So take it how you will.

John
=:->

>
> --
> Martin
>
>

-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 253 bytes
Desc: OpenPGP digital signature
Url : https://lists.ubuntu.com/archives/bazaar/attachments/20051021/4dacb79f/attachment.pgp