On Fri, Dec 11, 2015 at 04:47:16PM -0500, Michael McConville wrote: > Stefan Sperling wrote: > > On Fri, Dec 11, 2015 at 04:05:49PM -0500, Michael McConville wrote: > > > Bob Beck wrote: > > > > Stability before performance. Tmpfs does not have the former yet. > > > > > > ok mmcc@ for your PR_ZERO diff, as long as there's a comment added > > > about the performance impact and the potential to back out in the > > > future. > > > > I don't see the point of such a comment. > > Would you want such a comment everywhere PR_ZERO is used? What about > > calloc? Every committed change has the potential to be backed out for > > some reason. > > This just seems like a potential long-term, irreversible divergence. > Specifically, people may start making changes that rely on it or forego > NetBSD bugfixes related to it. IIUC, we pull tmpfs pretty directly from > NetBSD (where it originated), so I was trying to avoid that. Upon > further reflection, I don't know whether these sorts of changes have > already happened, or to what extent people care.
I don't think anyone can rely on comments you're proposing long term. IMO, just fix the code, and move on. Leave it to those who come after you to read the code you left behind properly. Yes, that includes reading code across function calls and verifying interdependencies.
