[mnet-devel] Twisted reactor vs. the DoQ

[Jim McCoy]

On Friday, July 4, 2003, at 02:15 PM, Zooko wrote:
> Okay, Twisted reactor and Twisted TCP *might* be faster than DoQ and
> TCPConnection on w32, as long as you use the compiled modules from 
> Twisted and
> not the Python code, but they might not, and they'll likely increase 
> the
> memory footprint a tad.

The benefit I see with twisted is that it becomes "someone else's 
problem" to keep that library working and bug-free. The twisted 
developers have a strong motivation to fix and improve twisted 
regardless of whether or not it is used in MNet, the same cannot really 
be said about the DoQ or TCPConnection.  That is not to say that there 
is anything critically wrong with these two modules that could not be 
fixed with a bit of effort, but I would rather save our time and effort 
for those problems which are specific to what we are doing. We are 
going to do this gradually to spread out the migration cost, but the 
long-term benefits seem to favor this choice.

A memory/speed hit in the short-term is worth the cost if it makes the 
event engine easier for us to use and extend. There are bigger fish to 
fry when it comes to memory usage. Most of the initial win32 
integration that is happening in HiveCache deals with the filesystem, 
the NT security system, and misc. things like the registry and NDS.

Jim



-------------------------------------------------------
This SF.Net email sponsored by: Free pre-built ASP.NET sites including
Data Reports, E-commerce, Portals, and Forums are available now.
Download today and enter to win an XBOX or Visual Studio .NET.
http://aspnet.click-url.com/go/psa00100006ave/direct;at.asp_061203_01/01
_______________________________________________
mnet-devel mailing list
mnet-devel at lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/mnet-devel