Back of the Bus Is Where the Cool Kids Are
How Microsoft Tries To Defeat Competitors at the Expense of Consumers
I just finished reading an article over at ‘I, Cringely’ about how Microsoft plans to use a new USB standard to cripple Linux. More appropriately, this article should is about how ultimately this is just going to cost consumers.
I say this because, as the article points out, people are going to go out and upgrade all their USB devices so that they aren’t limited to read only mode. But for those of us running Linux, well, I think we’ll likely keep what we have. Think I’m wrong? Well feel free to spend your money but I think I’ll keep saving up for that new Piper Seneca V that I want. The movie industry spent a considerable amount of money coming up with a protection scheme for DVDs and yet in 1999, people running Linux were able to run DVDs on their computers, without any licenced software and no matter what the region encoding was.
So while it might keep Linux down initially, I’ll be able to use these new USB devices eventually, if not right away.
Gentoo and stable ebuilds
I will start by presuming that you know about the Gentoo distribution of Linux and also about Gentoo’s ebuilds within the Portage package system. Let me also start by saying that I love Gentoo. I had been using RedHat since I started with Linux in 1997 migrated to Gentoo in September 2003 - and I’ve never looked back.
But the problem with Gentoo’s ebuilds right now is that they are inconsistant. For example, an ebuild for the new Mozilla Thunderbird 0.8 came out within 24 hours of the release. FlightGear 0.9.5 on the other hand was released July 29, 2004 and we’re still waiting for an ebuild. And for that matter, the ebuild for FlightGear 0.9.4 is still marked as unstable. These are two fairly extreme examples but you probably get the point.
Gentoo revolves about a group of volunteers, who like everyone else, has other things to do that more often than not take priority over Gentoo things. I’m not currently involved with Gentoo as a volunteer so some of what I am about to suggest may already be in place or simply may not be possible. But that said, here are my solutions to these things that I perceive to be problems.
I think that there enough people (who aren’t already on the Gentoo team) that can create ebuilds, that there should be no reason for there not to be an ebuild for the latest version of all existing packages with Portage. But, to maintain some stability obviously, ebuilds need to be reviewed. In a lot of cases, I think Gentoo developers cover multiple products and are busy with other aspects of their life. For that reason, I think that every package in Portage needs at least two people who share responsibility for that package, no matter how small or simple it is. In addition, every package should have a meta reviewer, a developer who is familiar and interested enough to ensure the ebuild is up to snuff and to add the ebuild to the CVS tree in the absense of the main two developers.
We also need a system within Gentoo to report that we have installed a package marked as stable and the degree to which it’s working, rated on a scale of 0 to 10, with 0 being it doesn’t work (and have reported a bug) and 10 being that it works perfectly and as expected. A system like this would serve several purposes; it would allow people to be notified when new unstable ebuilds are added to the system and it would allow people to see how others are fairing with the unstable builds.
In a lot of cases stable ebuilds appear within a few days of a package being released. This has resulted in me relying on using the portage system as a quasi replacement for Freshmeat and various project websites. I no longer needed to check for new releases as simply typing:
% emerge -upD world
would give me a list of what’s new. But, in some cases, this isn’t the case. To use FlightGear again as an example, version 0.9.4 of FlightGear was released on March 26, 2004. An ebuild for 0.9.4 arrived on April 2, 2004. That’s a reasonable delay, especially given that FlightGear relies on SimGear and in this case a new version of SimGear (and thus a new ebuild for that also) was also required. The bad part is that as of today 0.9.4 still isn’t marked as stable, and worse yet, 0.9.5 was released on July 29, 2004. And the worst of it all is that 0.9.6 is due to be released in the coming weeks. I built 0.9.4 on July 15th, when I happened to surf to flightgear.org to see when a new version was going to be released. Up until that point I had been running 0.9.3 and had no idea that 0.9.4 had even been released. Imagine my surprise to see that a new version of FlightGear had come out over 2 months prior. Despite the fact that the 0.9.4 ebuild was marked as unstable, I built 0.9.4 and have been running it with no problems whatsoever.
This is where that ebuild monitoring system would be handy. I envision this being set up similar to RedHat’s RetHat Network, where you registered each system and you could go to a website to see which systems needed what packages updated. That in itself would be a wonderful idea for Gentoo - instead of emerging each day so I could see what has changed I could simply log on to a website and see what packages that I have installed on my boxes have been updated. But more importantly, I would hopefully be able to see what unstable packages are available for my system, and what degree of success others are having with those packages. I’m not sure how best to implement a system like this but I’m going to look into it… along with the six b-zillion other ideas I have.
Thursday is blog day?
I haven’t gotten in the habit of blogging daily yet… partially because I only just installed Roller as a trial, partially because the CEO hasn’t approved opening this up to the whole world yet, and partially because I’m so busy with other stuff. However, for whatever reason I seem to always end up blogging on Thursdays.
How the OSS system is supposed to work
Ever since I updated my Mozilla Thunderbird from 0.6 to 0.7.1, I’ve lost the ability to sign and encrypt email messages. It wasn’t a huge deal to me since there are only 3 people that I use this functionality with, and was more irritating than anything. When I first noticed the problem, I first fiddled around with some settings on my machine as the upgrade had introduced a new version of Enigmail and I figured that perhaps I needed to update/change some settings; no luck. Next I checked the Gentoo bug database as more often than not my answer is there, given that I build everything using the portage/emerge system. Again, no luck. I then briefly checked the Mozilla bug database and didn’t find anything. At that point, I didn’t file a new bug because I prefer to be able to file helpful bug reports. Then, I forgot about the bug altogether.
Then, last week I discovered bug #6756 in the mozdev bugzilla. A bug report describing an experience similar to mine had been filed but the reporter wasn’t exactly clear and the developer was unable to replicate the problem. By this time, however, I had discovered how to replicate the problem and posted a note explaining in detail my set up and how I could replicate the problem. Five hours later, the QA contact for the bug posted a note that it was confirmed and a further two hours after that the developer assigned to the bug had found the bug and fixed it. Can things get much more brilliant than that?
So much to talk about, so little time
There is so much that I desire to write here, but there’s so very little time to do it in. And even less time is available because you really need to think things through before you put them down in print. Why? Because it’s ever so easy (and typical of human nature) to interpret something the wrong way or in a manner along the lines of how you are.
Setting up Roller
Roller has set a new record for length of time to set up a pain in the rear application; 7 hours of mucking around. Joy. I think a big part of the problem here is that there aren’t a lot of people using it with PostgreSQL. Combine that with complicated tools that are supposed to make development easier and you have an app that is impossible to work with if things don’t go smoothly on install. In fact, if it hadn’t just been chosen by Sun for their weblog software, I would have long since given up and moved on to something else. Fun fun.
