"Not Invented Here"
When I am hacking on Frisky I generally just plow though what ever I can because I have so little time to think about what I am doing. So when I have a problem say caching I just write some code to solve the problem not thinking about trying to use another solution. Then when I come aware of a solution there are two reason I would avoid using a third party. The first momentum / laziness which is pretty lame but with the time crunch it is what it is. Second reason generally stems from the solution to the problem I have written is a more correct solution than an external one. That is fundamentally the balance that is hard to figure what side of the line I should be on.
It not as bad as I thought, it is better!
Enough of that though really! It is time to use more robust and faster implementation of what I could write myself. Even if I had all the time the libraries make a lot more sense from a design and simplicity than to invent a very custom solution. Now that I have read up on both of these I just need to implement them.
Beaker caching rocks!
It is amazing that it took a recent blog post to finally make the bell go off that I should be using beaker. I have only heard Mark Ramm talk about it a couple times plus the TurboGears and Pylons mailing list talked about it a lot. Duh! Hopefully this will be quick and easy to implement and save a lot of time.
Message Queue with STOMP
When I first started to think about building Frisky I wanted a messaging system. I described it as a worker process and really detailed it out to be anti-cron but what I was getting at was a message queue. Ofcourse not until I was reading about message queue system in the latest python magazine did I realize this. This will take some experimentation but greatly reduce the code that still needs to be written to implement original feature set. I was thinking I may be able to programmatically start one of the Python STOMP servers that way it could just be a server setting to turn it on.
0 comments:
Post a Comment