February 2005 Meeting
Thursday, Feburary 9th at 7pm, UptownEspressoBelltownLocation.
Attendance
- (forgot name)
- (forgot name)
- Grissles (?), database work, 2 years Python for a living, email responders, interested in frameworks, playing with dot
Tim, last at a meeting last Summer, in Ballard, writing program to keep track of references, far enough to know how to start over
interested in TurboGears JohnDugan, been to other side of the lake before, worked for U of Illinois, did testing framework work, used Quixote to make web sites
- Yang, first time here, understanding what this is about, knows a little Python
LionKimbro (recording)
- Miles, plays with Python from time to time
- Brian, mostly database guy, db's significantly smaller ~ 4 MBs, not sure what's in there :), over 15 TB of (files) + more of (files,) much of it redundant
Talk
We had several demos:
YAPGVB Yet Another Python Graphviz Binding
"I've been using graphviz for a long - don't like the bindings and the interfaces - and wanted to learn Boost Python - so yesterday put on SourceForge pre-alpha release, single line of documentation: Fixme"
- What is Boost Python? It's like SWIG. C++ metaprogramming, to wrap Python. Can make a C++ class into a Python class. Hard to set up, but then makes wrappers so incredibly easy. Boost going into C++ standard library.
Ties into Python graphviz libraries. (?) Boost Graphviz library -- very powerful, easy to use, to talk with GraphViz.
Project name: ION:
- like ratpoison
- someone wrote a factoring program
- someone wrote a Federation and Empire play environment - no web page, can't be released
Where do we meet next time?
- next time, we'll do it at a house
- Lonnie offers his home in Capital Hill, John in Wallingford too, (has a projector)
Brian talked about a program he wrote:
- 10-12 physical locations with computers, some files need to be in various different places, a database knows what's supposed to be where. This is something so you don't have to do ad hoc cheesy script based manipulations with rsynch and stuff.
- Windows service sits on server, looks to see what should be there, works with a queue of what files to send where.
- Pieces are really dumb. Just get "this is supposed to go there. Delete this." blah blah blah
- With a queue, and you can move things around in the queue. (did I get that right?)
- Files are being generated all the time. (50-100 1GB files, every day.) 100's of small files, under 10 MB. Some are "REALLY BIG," bigger than the big files. Small file transfers are all automated.
- Queue exposed by a web interface.
- Database distributed; It's in a SQL server.
- Most files come in from one place, but they can come in from multiple locations.
- All VPN'ed.
- Two-way replication.
At this point, LionKimbro had to leave, so there are no more notes here.