[Stackless] Res: Stackless Digest, Vol 84, Issue 3

[Estevo]

> Hey Esteban,

>
> I can check this change in if there is no response, but honestly, that
> looked ghastly before your change and it looks at least as ghastly
> after :-)  In an ideal world this libevent-based socket module would
> be rewritten to properly support the socket module expected use
> patterns.

Yes, that bit was commented as a "XXX stupid workaround" and I didn't try to fix 
that (for one thing, I know nothing about libevent). The reason the other buglet 
came to my attention was because the socket management tasklet was keeping my 
program alive after my own tasklets ended.

By the way, the reason why I checked out this version of the stackless socket 
was because the standard one wasn't reporting any socket exceptions to me. 
Instead it would exit silently (raising TaskletExit, which is not normally 
reported) if, for example, if the server refused my connection. This is in 
Python 2.6. If this is not a known issue, I'll write a more detailed 
reproduction description.

Since the libevent socket is working fine for me and I'm only toying so far, I 
got lazy. If I stumble into any further misbehaviour I may look more seriously 
into either version of the module.

BTW, nice work on rpc.py. Much simpler stuff than I remembered from the days of 
proto1.

Regards,