Message141185
| Author |
pitrou |
| Recipients |
giampaolo.rodola, neologix, petri.lehtinen, pitrou, santoso.wijaya, socketpair |
| Date |
2011年07月26日.21:34:27 |
| SpamBayes Score |
0.00034955872 |
| Marked as misclassified |
No |
| Message-id |
<1311715978.8406.3.camel@localhost.localdomain> |
| In-reply-to |
<CAH_1eM3ji+ecQvF+3Y0y+6pg-rp6PNANBEDk-OEgr_AiY5VjDg@mail.gmail.com> |
| Content |
> When you join a thread which hasn't been started, you get an
> exception, not a deadlock.
But we don't have a deadlock here: another thread could perfectly well
call serve_forever() in the meantime and shutdown() should then return
after having stopped the event loop.
My point is that changing semantics may break existing software, without
there being a clear benefit in doing so. |
|