Wednesday, November 17, 2010

Manaiakalani - not all roses

I was writing a blog post this afternoon when I should have been working.

With the Manaiakalani project, I encountered my first major issue. And the bit I'm slapping my forehead about is that I had indicated previously that there may be a problem there but I wasn't entirely convinced that it was quite a problem.

And the problem? As a result of only really having one day for testing, and a day for sorting out the problems and a relatively small sample group (I had three laptops to test with and there was another laptop in the wild doing some testing outside of my efforts), and my dismissing of the issue, because we were testing at a pub and the pub's access point quite often has issues and the fact that very few people have been complaining about the particular issue on the Internet, lead me to not to look for a solution for something that I wasn't sure was a problem.

Sure enough, by around 10:00 this morning, I was called in to diagnose the problem. After a few hours of checking the DHCP server and the access points and trying to convince myself that it was a network issue rather than an issue with the image, I was writing up the error report.

And then it occurred to me that I had encountered this before. In fact, it was very much what I had encountered in the weekend and the error report was sounding very much ... exactly like what I had written in the weekend.

Suddenly my self assured "it's a network issue for the network team to fix" confidence was right out the window.

So I gave someone an external usb wireless adapter to try. Sure enough, no issues. So problem confirmed.

Now to find a solution... and quickly.

So the drivers are tied into the kernel. Included. Brilliant. Only, that would imply that to update the drivers I would have to update the kernel. I don't want to have to maintain a distribution. I'm there to make the distribution work for the user, not to replace the major bits of the distribution.

Consider that solution as a last resort. I won't do it. I don't want to. Imagine me stamping my foot. Yep - my foot's down.

So what are my options? Try and find an alternative way of doing it (without having to backport because I'm really just not that low level yet).

While downloading the sources for the drivers, I decided to try the ndiswrapper way of doing things. It works brilliantly. A half hour ping test and it worked flawlessy. Not a single lost packet. I'm loathe to use Windows drivers. It's almost insulting. Linux can get this right. However, if Ubuntu aren't willing to release a more updated kernel for Lucid, I'm pretty much stuck. I stopped trying to download the sources for the drivers - they were going to take near on an hour on a connection that has me drooling. In other words, loads of data for a small driver.

Unacceptable.

So now I've got to check the licensing on the Windows drivers, see if I can package it up, if I can't, install it individually (not acceptable for next years roll out which is 2000 laptops in 7 schools) or go back to doing the bit that I'm putting my foot down on.

Meanwhile, I think given enough time I might be able to figure out the sound problems. In other words, time in the middle of the night to sort this out.

In the meantime, while I was anticipating working tonight to package up those drivers, I've left the netbook at school. Whoops. I'll blame someone who isn't me.

No comments:

Post a Comment