Ingo Schwarze <schwa...@usta.de> wrote: > Hi Theo, > > Theo de Raadt wrote on Fri, Oct 30, 2020 at 12:10:41PM -0600: > > > Yes, that diff is a whole bunch of TOCTUO. > > > > If this was going to be changed, it should be in the kernel. > > > > But I don't know if it should be changed yet, which is why I asked > > a bunch of questions. > > > > Stepping back to the man page change, we could decide that accton > > should continue to behave how it does, and this conversation started > > because the man page fell into the trap of documenting the rc scripts > > RATHER than documenting accton. > > Given that nobody seems in a rush to patch the kernel, i suggest to > improve the accton(8) manual page for now. In particular, that > manual page lacks the required EXIT STATUS section, which happens > to be a natural place for mentioning what happens if the file does > not exist.
I guess so. > As usual, an EXAMPLES section is not strictly required, but in this > case, it seems useful to me, to save people from having to inspect there are only two ways to use this command. I don't think making one of them an example helps anymore. You have to be really dumb to misuse the command. I think this conversation came from a misguided origin. I think the man page documents accton correctly. The error occurs when it tries to document rc. But even that text is correct, it says: To have accton enabled at boot time Well if you don't reboot, it isn't enabled NOW, or it would have said "to enable and start accounting". I think this is making a mountain out of a mole hill.