On 8/24/22 16:27, Felipe Contreras wrote:
*If* my proposed patches are applied, SAVEHIST by default would be 1000, *and* if HISTFILE had some default--which is something my proposed patches don't do, but could be considered for zsh 6.0, *then* this could happen: 1. The user has configured SAVEHIST to something greater than 1000 2. The user starts zsh with NO_RCS (e.g. zsh -f) Yes, that is an issue, but I would guess those two conditions would only be met 0.01% of the time at best, which is why I claim this is a corner-case. Moreover, this could be easily avoided by HISTFILE not having a default if NO_RCS is set, could it not?
You keep throwing out these random percentages from no where, i am telling you as someone that helps people use unix shells over the years that starting your shell without your dotfiles isn't a corner case, there isn't a reason to guess, you can search this mailing list for the frequency of use, the README when discussing reporting bugs mention reproducing possible bugs from zsh -f. I've commonly seen people wonder why they lost shell history in other shells only to find out they started them in such a way their dotfiles wasn't sourced, when they increased the default size of history events being saved. But yes, in a hypothetical proposal of setting HISTFILE by default. not doing so when NO_RCS/-f is used would prevent the subtle data loss problem. Had i not chimed in on it though, i doubt the precaution would've been taken.