pushy is a nice project. i also thought of making a zero-install server, but
i realized services are a better direction. it allows you to use it over the
web, instead of the oh-so-lame currently used protocols, such as XMLRPC,
SOAP, WSDL, and what not. i don't want to pass large XMLs over an HTTP
server, just to invoke a line of code.

that was the purpose of rpyc. using a zero-install server basically means
going back to the classic rpyc (2.6),
where you have no control over what your client gets. it's a different use
case, and rpyc 3.xx supports it
as well.

if you want a "zero-install" server a la pushy, you can achieve by doing
ssh mymachine ./rpyc_classic.py -p 12345
... run python client ...
ssh mymachine pkill -2 rpyc_classic.py

so yeah, pushy integrates it all better and supports more transports than
ssh alone, but i don't consider
it as "extra power". you can write a small wrapper on top of rpyc, call it
`rpycssh`, that does the same.

as for the speed, i'll try to understand your code better when i have some
free time.
i don't see a reason that running a loop over 39 items should take 3.2
seconds, on the localhost.
i'll look into it.


-tomer

An NCO and a Gentleman


On Thu, Mar 10, 2011 at 07:38, Fruch <[email protected]> wrote:

> here is the code:
> https://gist.github.com/863616
>
> (you'll need to start a rpyc server and pushy server on localhost in the
> background,
> mine was embedded inside a C program, so I didn't attached it)
>
> from my POV, pushy is actually achieving all the three goals you've
> mentioned.
> and as he said on his blog, RPyC was his inspiration.
>
> he sure did a good job. I still think we should reconsider merging with his
> code.
> adding a wrapper to support the old rpyc code
>
> BTW, I've found this:
>
> http://stackoverflow.com/questions/1239035/asynchronous-method-call-in-python
>
> it has in one of the replays nice replacement for async, that should work
> with client code only (I've haven't tried it yet with pushy/rpyc)
>
>

Reply via email to