Rui CarmoSS7 training is over and done with, and umpteen work e-mails have been duly read, prioritized, and processed. My Windows domain password is due to expire sometime tomorrow, and my only hope right now is that it doesn't happen before 9AM (David has parodied IT policies, but despite the hassle, periodic password expiry is one I fully agree with).
So, to relax a bit, I've been working through my long list of mini-projects:
- Tried to get gtkmozembed to return the number of bytes downloaded for any given page (the whole thing, images, plugins and whatnot). The initial idea was to pick up this thumbnail generator and use it for measuring page load times and volume of data. Page load times are trivial, but the progress_all signal just can't be working properly in Mozilla 1.7.3 - not when it fails to report 90KB worth of image data. No, it's not Mac coding - it looks like I can do it trivially using Safari and AppleScript (with the added bonus that I get anything requested by plugins), but this has to run on a Linux box. And no, I don't want to use khtml.
- newspipe has been acting up on me again - since I have a monstruous feed list, I am getting repeated items in some feeds (completely identical, repeated items, down to the checksum headers). But before I build a testbed to work on this, I am slowly merging the last CVS changes into bayespipe to see if I finally get it going again. Some ideas floating around my head are: breaking up the pickle files into per-feed caches and using a lot more locks and building a set of classes to abstract everything related to item storage and making sure those are both thread-safe and multi-instance-safe (we need something with transactional commits, I think).
- I've been meaning to do the simple Subversion test that Melo mentioned for ages now. Then I realized I'm not grokking Subversion branching at all, and that I need to read the book again.