Re: [SLE] SuSE 8.2 -- What is the best file system?
Hi Anders,
So, is ext3/ext2 the standard file system for GNU/linux?
I don't believe you can speak of a standard. The "vanilla" kernel >has several file systems, and they're all interoperable, in the sense >that a file stored on one system can be moved to another without alteration.
That makes sense to me.
If so, why spend so much time developing other kinds of filesystems like Reiser FS, XFS, JFS?
One file system is good at one thing, another are good at different things. Hans Reiser & co. have very specific ideas on how to develop >a file system, the people at SGI have different ideas and their experiences with high performance graphics systems has evolved into present day XFS.
This is indeed true. However how does this help a "normal" home PC user decide on a file system? I don't run high performance graphics programs on my pc. Does this mean I definitely should not use XFS?
A file system is critical to any operating system and I think it would be much better if everyone could just focus their efforts on improving one file system. This is the case with MS Windows. They started out with FAT, then moved on to FAT32 and finally are using NTFS. The same goes for other stuff in the open source community like KDE and Gnome for example. This "multiplicity" approach hurts linux as a whole. Linux software as a whole should not compete against other linux software but should compete against other OS's software.
Head down to your local library and check out a book called "On the origin of species" by Charles Darwin. It may not be on the comp.sci. curriculum yet, but it should be.
This is certainly true for living organisms. I think software falls into a different category. Focusing all attention on one project instead of 4 or 5 would be much better for the software community in the long run. I mean what is the difference for a "normal" linux user between XFS and JFS? ~~Nick _______________________________________________ Join Excite! - http://www.excite.com The most personalized portal on the Web!
On April 13, 2003 03:29 pm, Nicholas Parsons wrote:
This is certainly true for living organisms. I think software falls into a different category. Focusing all attention on one project instead of 4 or 5 would be much better for the software community in the long run. I mean what
It's great until some time in the future you find out the choice made was the wrong one. It will happen. It has happened. Nick
On Sun, 2003-04-13 at 21:29, Nicholas Parsons wrote:
This is indeed true. However how does this help a "normal" home PC user decide on a file system?
If by "normal" you mean a person without much computer knowledge, someone who isn't able to digest the technical differences between the file systems, then I'd guess most "normal" users wouldn't decide on a file system at all. They go with the default. With SuSE that's reiserfs, with Red Hat I believe it's ext3. They're both good enough for most intents and purposes
I don't run high performance graphics programs on my pc. Does this mean I definitely should not use XFS?
Not at all. It just means you're not part of the intended market for which it was developed. You can use it anyway and get very good results from it. I do.
Head down to your local library and check out a book called "On the origin of species" by Charles Darwin. It may not be on the comp.sci. curriculum yet, but it should be.
This is certainly true for living organisms. I think software falls into a different category.
Only in the technical details. I believe the priciples still apply. Lots of projects living and thriving will produce lots of different ideas. The best of these ideas will over time evolve, possibly into still new projects, each better than the ones that came before it. It's just the nature of the beast. One single project only leaves room for one single set of ideas. Imagine yourself waking up in the middle of the night with a brilliant new idea for a file system that would give 10 times better performance than any other for one specific purpose. Should you not be allowed to pursue that idea? Should you still be forced to solve bug #xxxxx in ext3's bugzilla. That's not the way forward. Maintenance of existing projects is needed and vital, but so is new projects with new ideas.
Focusing all attention on one project instead of 4 or 5 would be much better for the software community in the long run.
No, in fact I think it's just the opposite. It may be better in the short run, but in the long run I think it's the way to certain death.
I mean what is the difference for a "normal" linux user between XFS and JFS?
With the above definition of "normal", none. All file systems work the same way, they all use the same commands for opening, writing and reading files. But some day the "normal" user may take the next step, and find that there are indeed subtle differences that - for a more advanced user - gives one an advantage over the other. When that day comes, he'll be glad he has the choice.
participants (3)
-
Anders Johansson
-
Nicholas Parsons
-
Nick Zentena