Gotta love "notwithstanding"
Posted Dec 7, 2012 13:12 UTC (Fri) by man_ls
Parent article: A FALLOC_FL_NO_HIDE_STALE followup
I thought that the technical reason was that the flag was useful only for one filesystem (ext4), and only because it had bad performance when returning unallocated blocks because it had to zero them first. In the original article it was claimed that the right solution for this problem was to fix ext4, not hide the problem with a flag. Looks like sound technical reasons to me, Linus trash talk notwithstanding.
There is also something slightly annoying when a process gets subverted by a maintainer just because the company paying his bills needs it. No technical excellence to be seen anywhere. Perhaps my ignorance about this problem is even deeper than for kernel development in general, kind of a Marianas Trench of knowledge.
to post comments)