Fixing Linux, but does it really need that much fixing?
2 December 2008
by
Gunthy
Filed under
Linux
Over at informationweek.com one of their authors has posted an article about what needs fixing in Linux. While I’m not going to repeat what they said there (I’m sure you all can quite read it for yourselves). I do want to discuss a few things about it here.
It seems that this writer does not really have that much in-depth knowledge about what is going on in the FOSS world, and how this community works. This is also being reflected in the comments on the article. It looks like this person has fired up one or a couple of popular distributions for a limited amount of time, but didn’t really like what he saw, writing this twisted article as a consequence.
However it is true that Linux, or any Unix for that matter, still has quite the long way to go. While some in the Unix-family surely are more mature than others, especially when you take a look at proprietary Unices, I hardly agree with what was written.
A lot of the article, in multiple of it’s points, is in essence about the fact that things are so de-centralized. Well what do you want? We are talking about Open-Source here. Most of the work is done by volunteer-programmers that devote their free time to all these projects. There is no central leadership and while some of the bigger projects have showed that they do posses quite good management, most of them are just people having fun.
The first point discussed is the different types of package management. True there is not really a standard, although the RPM-format is already widely accepted, as is the DEB one. In fact, these two alone are probably the most used in Linux-land, and there do exist tools to convert one to the other without much hassle.
Second up is the de-centralisation and text-based configuration files. With the first part I totally disagree. Apart from some exotic software here and there, all of the config-files are stored in the /etc folder. The fact that it’s using text-based files actually poses an advantage to me. It’s easier to backup, and you don’t have to spend endless amounts of time browsing through one big configuration database. I know other OS-es do it differently, but if you like that method, then use that OS!
Point three is up. The author is discussing the problems with the kernel Application Binary Interfaces. Big words, and although I’m no expert on this one myself, I don’t really see the problem here. While it’s true that ABI’s are in charge of interfacing between the kernel and applications, these days most applications use higher-level libraries. So unless you are using a lot of proprietary drivers or so, this really is not such a big deal…
Up next is Native File Versioning. Basically, this is functionality that allows you to restore a certain file or folder to a previous version. The author seems to be used to this from his Windows experience, but being a Microsoft connaisseur myself, this really doesn’t work that well anyway. Admitted in some cases it can be handy, but I haven’t used or needed this even once in the many years I’ve been using computers. If you’re smart and you’re going to edit critical files, for god’s sake, make a backup of the file before you start…
On the next two points in the article the writer discusses the Graphical User Interface and X11 integration of applications. I’m going to vent my comment on these two together, as in my opinion they are quite linked together. One of the things I’d like to say about this is that the proposed features by the author really don’t seem to add any value to me. Integrating the GUI more with the kernel only provides semi-lock-in scenarios which should be avoided as much as possible. The GUI’s are already having a lot of standards, more than enough it seems to me. Secondly the article discusses the problematic behaviour that when your X crashes, all applications running under X crash as well. When reading this I was kinda speechless. I mean, isn’t that kind of logic. Is restarting X so difficult? No! Even the author admits that… Actually, the times current X sessions crash isn’t that often anymore anyway, so why even bother on this one? I’m sure over time more and more stability will become available.
Last but not least is the ‘lack of’ commercial backup utilities. Again I totally disagree on this one. All the big players on the backup market already provide full support for almost any Unix/Linux flavor. In fact, most of these things actually work better in Linux, at least in my experience. For end-users the options might be a bit more limited, but if you look around a bit harder, you would find plenty of options, and if you insist on using something ‘Windows-only’, you can always try to run it in Wine, emulating the Windows environment.
Conclusion, this article really dropped the ball on this one. I really am not the type to start writing huge comments on other people’s articles normally, but this one made me feel I should make an exception, and so I did.
