Slashdot Mirror
Rasterman Responds To Seth And Havoc
An anonymous reader writes "Rasterman, of Enlightenment fame, has responded to Seth Nickell and Havoc Pennington's blog entries, which were in reference to this previous article. about Next gen X rendering. Raster says: 'Well it seems the XDevConf has produced some interesting blogs and discussion. I'm a bit sad I was not able to attend (no funding at all), as there seems to have begin a lot of discussion and moves in directions we in Enlightenment land have been going for years, and are likely far ahead in. I guess this means we haven't been able to share our experience in this. Maybe next year. Anyway the point is that this has started up some musings from Seth Nickell and Havoc Pennington related to this. This is great - finally people are beginning to take seriously what the Enlightenment crowd have been talking about for years.'" (Note: the previous post was about Nickell's post, not the other way around.)
I can see where he is coming from, but for all the hype the E team generate over their amazing new libraries, how many apps actually use them? As far as I can tell, basically none. I don't know why that is though.
If you choose to write open source code, you are chosing to have no money. That's your choice. But dont complain about it.
Open source is not an end itself. With some celebrity exceptions, open source exists because someone solved a problem--often a business problem--and released the solution to the public.
Why would they give it away? Because they have no interest in trying to sell it. Selling shrinkwrap software is a tough business, most people would rather focus on whatever it is they're better at. They stand to gain much more by open sourcing it than they would keeping it in a vault, or trying to sell it.
Frankly, I don't think there _is_ a speed issue with X11. There are performance issues on XFree86/XOrg with some (many) drivers, AFAIK mostly due to limited developer time and limited access to hardware. The fact that the current software RENDER implementation is not signficantly optimised, and few drivers implement RENDER hardware acceleration, does not help.
Working on my NVidia equpped box here (GeForce Ti, nvidia drivers, but for 2D 'nv' is almost as good) X is much snappier than I usually find WinXP to be. Turning on RENDER acceleration has helped a lot.
I'm sure folks will bring up the "because of the network" myth up here, so let's get this straight - any slowness in X is not because of network support. Go ask Keith Packard, I'm pretty sure he's been rather clear on the matter more than once. My personal, very much non-expert understanding is that most performance issues peope experience are due to limited hardware acceleration and inferior drivers.
If you don't believe me about how much difference the hardware and drivers make, go find an S3 based system, preferably S3 Trio32/S3 Trio64, and compare it to a PCI-based (to keep it fair) NVidia GeForce 4 MX on the same hardware. It's like they're two totally different computers - the change is jaw-dropping. I use thin clients a lot, so I care strongly about video performance and tend to notice these things.
It's also worth noting that hopefully many of these plans will lead indirectly to performance improvements, by making RENDER acceleration and RENDER optimisation pretty much mandatory.
Last Fall, I had a serious focus bug in Enlightenment (e16) that would lock my mouse to a particular region of the screen and require X server restarts. It would usually happen at the worst time, when I was working fast (busy!).
I worked with e-team member Kim Woelders on the problem and he produced a couple of patches after I sent him some good reproducible test cases. We exchanged a total of 39 email messages and it was finally fixed. I'd usually have a patch within 24 hours of sending him a test case.
All of that while they are busy trying to get e17 out. The work that the team does is amazing and I am very grateful.
To say that I am a fan is an understatement!
Anecdotes != Data
I can easily counter that on my dual-boot system with Windows XP, Fedora, and a tweaked Gentoo, both Linux distro's are far, far more "sluggish" than Windows is. Oddly enough, what gives Windows a real kick in responsiveness is the I/O subsystem. Running Windows 2000 on a 400Mhz PII laptop was dog slow. Running Windows XP on a 400Mhz PII with SCSI RAID underneath and it flies. Linux/X11 does not on the same hardware, regardless of optimizations, distros, windowing managers, etc. I use this largely as a plaything, and as such have played with a LOT of distros, tweaks, and window managers over time.
So are you right? Am I right? We don't know. Does anyone have *real* data or studies on this, or just a bunch of anecdotes?
I don't know what kind of crack I was on, but I suspect it was decaf.
X. Is not. Network Based.
Not on the local machine. For local displays it doesn't use any networking at all. It uses UNIX pipes which are very fast and also DRI (Direct Rendering Interface) to talk directly to the video hardware.
I wish this myth would disappear. X only uses networking when using it over a network.
University - a box of academia nuts.
First of all, did I just read a story that gave all the background to Rasterman's response, but left out the actual response itself? Nice.
I've always liked the Enlightenment project, and I try to keep up-to-date with what's going on (which is not easy), but it seems pretty clear to me that it will not be the future of the Linux desktop.
E is not really a valid option for the OSS world - I wouldn't be surprised if more people were using XFce or Rox than E at this point. Sure, Linux itself has proven that if something new and amazing comes around and blows everything away by a large margin it may have a *hope* of shifting the momentum, but as great as E is, I doubt it is that impressive.
The reason why the framework Seth+Havoc describe will win over the E stack is because it is integrative, whereas E is not. When the next-generation X rendering system is in place, it will be available to everyone who uses those extensions. Probably by the time Damage + Composite are enabled by default on X, the latest KDE + gnome desktops will have support for them. And all the applications in those respective desktops will quickly (if not instantly) gain those advantages. Remember when the same thing happened with anti-aliased fonts a few years ago?
Yes, you can get the E magic right now, but you have to go through E. As long as they remain the sole gatekeepers, you can expect them to have the same extremely limited influence they have now. At this point in the game, I seriously doubt they can beat the inertia from the other desktops. Honestly, if you're developing a new application, are you going to develop for the mature and distributed kde or gnome desktop environments, or will you use E, which is available now with some ephemeral advantages but some serious disadvantages?
It's also true that by using E you're not committed to using _only_ E, but then, what's the point? If you use E + some random GTK application, you're not going to get the consistent graphical features until GTK itself gets those features... but at that point all gnome applications will have them.
The example of the Cathedral and the Bazaar is a good metaphor for these differing stacks. It seems to me the E project has always been fiercly exclusive in the way it does things - the whole Enlightenment Foundation Libraries are the best example of reinventing the wheel with E technologies. But the cost they've paid is limited deployment, slower releases, less interest and a rather narrow development strategy. Certainly that may suite some people fine. However, with that in mind I don't know how reasonable it is for Raster to be calling sour grapes.
http://www.talknerdy.org