On Fri, Jul 12, 2019 at 08:46:19AM -0400, Greg Wooledge wrote: > On Fri, Jul 12, 2019 at 11:35:20AM +0300, Reco wrote: > > I don't dispute that RedHat did a lot of good things - good chunks of > > the libc, gcc and a kernel itself is wrote by them. > > On the other side though we have some really controversial things like > > SecureBoot support, Wayland, GTK3, xfs, and s*****d. > > XFS was developed by SGI, not by Red Hat.
Yet it's their favorite toy now, and have been for at least for the last decade. And "problematic" does not even start to describe this particular filesystem as of now. But I'm an optimist. Give it ten more years and maybe something good will come out of it. > There's some stuff about Red Hat in subsequent paragraphs, and it > definitely does not paint a pretty picture of Red Hat. If wikipedia is > to be believed, Red Hat added partial XFS support 7 years after Gentoo, > and then charged its customers extra money for the basic command line > tools. (And then possibly stopped being evil a few years after that. > It's a little unclear.) I was not aware of this little tidbit, but it sounds like very thing RedHat would do. For me it was enough that they made xfs the default one (some can say "forced", but note that I didn't say it) and they *knew* that xfs will lead to data loss if used without battery-backed storage. > I'm not sure exactly what you consider "controversial" about XFS. It's > just a file system that you can choose to use, or not. Random slowdowns for no good reason. Data loss on power failure. Kernel panics at xfs-specific parts of the kernel just because. Saw a lot of such stuff. The solution was the same every time - fsck it (ambiguity is intentional here), we're moving survived data to ext4. Now, ext4 is not without its problems too. But it's the scale of such problems that matters. Reco
