I’m sitting in Stewart airport at the moment, happily connected to ye olde interweb over my Sprint PCS phone. While this might not be exciting to most folks, this is the first time I’ve gotten around to trying to make this work in the year since I’ve had the phone. There are pretty good instructions for getting this to work on an older sanyo model, and sprint, and other than moving a couple files around to match how Mandriva lays them out vs. Debian, everything was good.
Well, at least it was after I remembered to tell my firewall about ppp interfaces, as it was dropping all my packets because they were on an unclassified network. 😉
Long airport waits are going to be a heck of a lot better now. I’m pretty psyched!
I had an interesting experience this weekend, in which I was finishing up my CS project for the semester (an Assembler written in Java), and found that I didn’t really hate Java that much. The API is still a bit odd at times, but since they added basic regex support in 1.4, it isn’t nearly as painful to use for jobs that would be trivial in Perl or Python. I think that after 3 classes in a row with Java projects at the end of them, I’ve finally learned how to keep the language from getting in my way. And of late it is also the only reasonable sized programming project that I’ve actually had time to sink my teeth into, and actually getting things accomplished always colors things in a more positive light.
More and more IRC has has basically become a dashboard for me. Most of what I need to know goes on in it, and most of the technical conversations I have are based in there. Especially given that the 2 main tools I use, mercurial and moinmoin, have rss feeds, it becomes even more sensible to have that information feed right into IRC as well.
So after spending a couple hours understanding POE (which is an event model for Perl much like Twisted is for Python), and the IRC and RSS sample code that goes along with it, I created a simple rssbot for the mhvlug IRC channel. It’s mostly working at this point, though I’m going to clean it up a bit and add a config file over the next few days before publishing it.
POE is definitely a much better place to start experimenting with IRC bits that I was thinking about before. And it was fun to write an IRC bot from scratch in 3 hours. 🙂
I finally decided that I had enough of Trac, the integrated wiki / tracker / scm site, after it appears there are no longer any working versions in svn that support mercurial as the back end source management system. That wasn’t the only reason that I decided to leave trac behind, it was just the straw that broke the camel’s back. I know if I pinged the developers, it would probably be fixed in a short while, but after a month of Trac at work testing things out, I’ve decided that I just like MoinMoin better for a wiki, and the other benefits weren’t something I was really using.
There are a bunch of reasons for this:
- MoinMoin supports Templates for pages, this makes creating similar content pages (like interview forms for interns) much easier. In Trac I had to cut and past a lot.
- While it shouldn’t be a big deal, the double click page to edit preference is so nice from a usability point of view
- I was doing other MoinMoin work on an official wiki where it was convenient to have my own test wiki to toy with, write new Parsers and Macros
- The benefits of the RSS feed for Trac is heavily mitigated by Mercurial having it’s own RSS feed built in. I also found the RSS that Trac generated to be less than useful, and will probably even make some enhancements to the Mercurial feed as well.
- There were just enough format differences from Trac to MoinMoin, that it was frustrating me in moving from one to the other.
- I’m already maintaining MoinMoin for MHVLUG, so one less app to maintain myself is much better.
- MoinMoin 1.5 just feels slick. The built in styles get everything you don’t need out of the way, and the GUI editor is a good addition to get new people using the wiki.
So trac.dague.org has been retired, and wiki.dague.org is where I’ll be putting all my bits now.
I just pushed out version 0.2 of rpm2rss, which I need to announce on FreshMeat still.
This is an idea I came up with last month, as I maintain a number of Mandriva RPMs in a number of locations. Honestly, most of the time I just push them out there, and hope other people figure out what I pushed, but that’s a bad solution.
All this program does is take a couple of well known CPAN modules, RPM2, XML::RSS, and AppConfig (for config file parsing), and brings them all together, with a nice interface and man page. However, it means I can now run this under cron control on my systems, and not have to think about it any more. An example output feed can be found here. Given that I already have 65 URL hits, and 4 subscriptions to the project on FreshMeat, it seems like other people are interested in it as well.
I still need to get up the website listed in the man page of the package, and remerge in my tuxplus stuff to that, but that should be done today.
Over the weekend I finally got around to actually finishing the openwrt install on my Linksys wrt54gs that I first modified back in June. This means I finally have both 802.11g in the house, and I have traffic shaping on my outbound connection.
While not all the kinks are worked out yet, but moving the router from the living room to the office, and actually putting it in line with my upstream provider, was a huge step forward. Let’s see what my next long overdue project is to accomplish…
I got a little more progress done on TuxPlus over the weekend, my attempt to build a Linux recording program for the PlusDeck. I stopped being a complete retard, and started using the Device::SerialPort perl module for talking to the serial control port. From that I was able to map out all the RX serial codes from the PlusDeck, all mapped out in TuxPlus::Constants.
I should put together a sample control program, which might actually be enough to release the code. My next major challenge is actually interfacing with sound recording, and ecasound looks like the current best approach there. Though after realizing how bad things sound if the recorder isn’t running at realtime priority, I may just have to write a setuid wrapper with a very small communications interface to it.