Andrew Dunstan <andrew.duns...@2ndquadrant.com> writes: > On Sun, Jan 5, 2020 at 6:29 AM Tom Lane <t...@sss.pgh.pa.us> wrote: >> What has to have broken those machines was the change to explicitly >> force TERM to "xterm". Now I'm wondering what their prevailing >> setting was before that. Maybe it was undef, or some absolutely >> vanilla thing that prevents libtinfo from thinking it can use any >> escape sequences at all. I'm going to go find out, because if we >> can use that behavior globally, it'd be a heck of a lot safer >> solution than the path of dealing with escape sequences explicitly.
> You can see what settings it started with, although only certain > values are whitelisted. See orig_env in the config. e.g. crake, which > is now failing, has no TERM setting. Ah, right. The one-off patch I just pushed also confirms that the TAP tests are seeing TERM-not-set on Andres' machines. I'm currently investigating the possibility of just unsetting it in the tab-completion test and then being able to revert all the escape-sequence hocus-pocus. It's looking promising in local testing, although Peter said off-list that it didn't seem to work in his setup. regards, tom lane