I’m about to deploy Python to user boxes and farm nodes and wanted some feedback on what others have included in their deployments, keeping python updated, adding modules and any issues you’ve faced?
Our network is 72 machines, mostly on winXP32.
15 newer user boxes are on WinXP64.
Yeah, you probably don’t need both Pyro and Twisted. They’re both networking libraries, along with RPyC.
Unless you need every one of those for existing tools I’d suggest keeping it down to just what you need. As long as you have a method to push out new extensions later that is. But that’s completely personal taste.
One thing I’ve wanted to try is installing EasyInstall and use that as the method of deploying/updating extensions on studio machines. Rather than pushing the post-install files like we do today. It would need to handle failures when contacting the servers and retry until an update succeeded, but otherwise could be easy to manage that way. I’ve only used EasyInstall locally so far, however.
Working with my IT guy, his concern is that installing python on all the machines opens a pretty big security issue. It’s a programming framework with full access to just about everything.
I’m now looking at py2exe to create complete packages per script, but this will lead to alot of redundant information.
I assume it’s better to just have it on each system, but how are you dealing with the security issues?
I’m almost inclined to say that if they know what they’re doing with python, then not much your IT guy is going to be able to do will stop them, with or without python So its maybe more of a personnel issue?
If a user is logged in and working, they’re presumably permissioned enough to mess their own workstations up pretty badly. So access to Python present on their machines is really not making anything worse.
As long as non-priviledged employees are unable to distribute new scripts to the studio that run automatically, I’m not sure what the concern would be.
Here’s the list of extension packages we currently push to all devs here. We do this mainly so TAs can count on them being present on all machines, and not resort to py2exe, etc.