[MERGE] Fix bug #225020 by catching CURLE_SEND_ERROR error.

[Vincent Ladeuil]

>>>>> "robert" == Robert Collins <robertc at robertcollins.net> writes:

<snip/>

    robert> bb:approve

Thanks.

    robert> FWIW I think its a change in socket behaviour in the kernel, that also
    robert> made bzr's minor idiosyncracies in  1.5ish make hpss really slow (until
    robert> we fixed them). So likely a very old curl bug :).

Could be, when debug is activated, pycurl emits a very suspicious:

* HTTP 1.0, assume close after body

This one has puzzled me for a long time... I don't know what the
library really does in that case, but...

    robert> Consider filing a bug upstream too please.

Done.

        Vincent

P.S.: Thanks for keeping me informed about re-installing pycurl on pqm.