EXT3 Driver Might Not End Up Getting Removed From Linux 4.3

[ErikR]

I think that this shall take the same long and cautious procedure as new file systems does; new stuff is made available and the binary distros takes like forever to enable them in the default kernel. Like with BTRFS. Phasing out a file system should be similar, let the major distros diable the file system by default and see if there are any troubles with that. ETX3 can possible be enabled by additional kernel-option to emphasize the shift. Just like with systemd where you needed a flag in Arch Linux to enable it at first, then gradually the shit took place.

Can't really see why removing EXT3 form kernel can't be allowed to take some time, or even why it is important. For god sakes, there are tons of file systems with comparable only a few users still in kernel, why touch one of the important EXT systems that we have been relying on like forever.

I myself are using EXT3 and have compiled support for it into kernel. I am using gentoo and configures my own kernel and will most likely try to add EXT3 as an additional kernel module if gentoo maintainers makes it available, or wait a couple of kernels before upgrading as my system runs smooth and fine with 4.0.5 and ZFS is working just perfect.

[ErikR]

By the way, after 20 years of Unix at work and 10 years of Linux at home, mostly Gentoo, I still do not know what LKML is or which of the thousand forums I should debate EXT3 on.

[Zan Lynx]

Why not simply ask in user forums? Crowd sourced intelligence ftw

What they should really do is add a telemetry reporting package so that the kernel reports what everyone is using for hardware as well as a few randomly selected blocks of their RAM and hard drive contents. Keystroke and mouse data reporting will be optional.

[jpp650]

Bravo Linus !!!
A slap in the face of the developers dictatorship:

<<we develop this, so we make and change all what we want.
we don?t mind users feedback, installed base or bussiness.
if you don?t like it, develop yourself an alternative or maintain it by yourself>>

Really, sometimes I feel that I?m living under that dictatorship rule in examples like:

- gnome 2 -> gnome 3
- X-windows -> Wayland
- gcc -> llvm
- ext4 -> btrfs
- x86 (32b) -> /dev/null
- ext3 -> /dev/null
- SYSV init -> systemd

Some of the suppossed successors ( btrfs, gnome 3, systemd ) are a crappy piece of sh***it.
Linux as a project needs much more common sense forcing those transitions,
or it will not replace UNIX, Linux will be replaced by Windows Server XX-whatever.

[nils_]

I understand Linus' reasoning, however, I don't think it applies. Anybody who is still stuck using an EXT3 system probably isn't even past the Linux 3.0 kernel.

I agree. How many people are actually running the most recent Kernel, while not also using more recent filesystems? Hell some distros are still stuck on 2.6.32.

[mup4]

I would like to point out (as I am not subscribed to the LKML) that the current DCP (digital cinema package) specification is still "stuck" at ext2 mandatory support : https://en.wikipedia.org/wiki/Digital_Cinema_Package

ext2/ext3 support simply cannot be taken out of the kernel tree for this and many other reasons.

[geearf]

I think this quote explains it all

Both Red Hat and SuSE, as well as Debian and Ubuntu, are using ext4
with CONFIG_EXT4_USE_FOR_EXT23 for a couple of years now to support
ext2 and ext3 file systems. So with the exception of some really
ancient enterprise Linux distros, and people who are manually
configuring their systems, very few people are likely using ext3 code
base, which means the chances that it bitrots increases.

I've looked and it's enabled iin Arch as well.

[johnc]

I think some of those NAS boxes ship with ext3, like Synology or whatever. I could be wrong though.

[mup4]

Maybe but again to take the example of digital cinema DCP "readers" (called servers like the Doremi DCP-2000, the most widely used) are ancient and obscure distros.

[Scimmia]

Can't really see why removing EXT3 form kernel can't be allowed to take some time, or even why it is important. For god sakes, there are tons of file systems with comparable only a few users still in kernel, why touch one of the important EXT systems that we have been relying on like forever.

Bravo Linus !!!
A slap in the face of the developers dictatorship:

<<we develop this, so we make and change all what we want.
we don?t mind users feedback, installed base or bussiness.
if you don?t like it, develop yourself an alternative or maintain it by yourself>>

I would like to point out (as I am not subscribed to the LKML) that the current DCP (digital cinema package) specification is still "stuck" at ext2 mandatory support : https://en.wikipedia.org/wiki/Digital_Cinema_Package

ext2/ext3 support simply cannot be taken out of the kernel tree for this and many other reasons.

People, please stop and read what is actually going on. Nobody is talking about dropping support for ext2/ext3. They are talking about dropping the separate driver because the ext4 driver supports those previous filesystems, making the separate drivers redundant. This would have very little affect on anyone.