On Thu, Mar 02, 2023 at 07:18:41PM -0700, Theo de Raadt wrote: > What is with the long mail?
Theo: Daniel Dickman asked a question to me, on-list. I provided the answer, and tried to close the issue once and for all to avoid further replies. You replied to the thread. List: The truth is that Theo made a mistake when he replied to my original post, by assuming that this was a Python issue: > It is encouraging a large application to do a round-trip through the > kernel, for a rare occurance. Since he can't provide any sensible technical argument for not including the four lines to implement /dev/full, he's resorted to ranting. I don't think that anybody is impressed, least of all me. The intended use of full is for testing how simple shell scripts behave when they hit an unwritable file. > It should simply skip doing the round-trip through kernel. This comment is pure bullshit. Applicable to the python case and bascially nothing else. To test a shell script in this way you would have to remove each line that did file access one-by-one, and replace it with code that simulated a failure. Obviously ridiculous. > I see absolutely no point in wasting bytes on the root partition, or bytes > in the kernel, for this stupid idea. Seriously, this is a joke. One inode on the root fs, and a few bytes in the kernel, which could be recovered if somebody applied the diff that I posted to remove the /dev/io / 14 check which has been wasting bytes in OpenBSD for a over a decade, because none of the "developers" noticed it. This list, (and even more so -misc), is wasting too much of my time now. I'm taking a break. People know how to contact me off-list.
