Time for ext4?
Posted Jun 18, 2006 15:48 UTC (Sun) by malor
Parent article: Time for ext4?
The biggest reason a new, incompatible FS should be called ext4 is because of user ignorance.
Not all admins track lwn.net or the LKML. (I certainly don't touch LKML, sheesh!) There's a fundamental expectation that if it's ext3, it should be mountable by anything that supports ext3.
Calling it ext4 automatically will map it, in most people's heads, to the correct ideas:
1. "It's in the ext series, so it's probably stable."
2. "It has more features than ext3."
3. "My existing ext3 kernels will not be able to read an ext4 volume without an upgrade."
4. "I will be able to upgrade from ext3 to ext4, just as I could from 2 to 3."
If, on the other hand, you call it ext3, but then break the compatibility, I guaran-damn-tee you there are going to be frustrated, unhappy users out there. They're gonna be PISSED when they can't fall back to an earlier kernel, or when they can't copy files off their thumb drive. If you call it ext3, you're making a guarantee about portability, and breaking that WILL cause a lot of user frustration.
The ext guys are right that they have huge street cred. They have so much cred, in fact, that calling it ext4 will improve uptake. People trust ext. I trust ext. I'd happily run ext4 the very first time it was available.
But forcing invisible upgrades is a _bad idea_. Even if the new ext3 features require special command-line arguments, at least some distros are going to use them, and at least some users are going to be bitten.
Wanting to call it ext3 is a feature to benefit the developers; they want their code to be as widely used as possible. It is NOT a feature that benefits users.
Be fair to the user base and call an incompatible upgrade an UPGRADE, not the same thing.
to post comments)