On Fri, Mar 27, 2026 at 4:20 PM Jacob Champion <[email protected]> wrote: > I don't think I can enforce either choice, though: I pass the > error_detail into the ereport(FATAL), so the process is about to go > down, and I'm never going to pfree() it.
(I also think it's reasonable to do something like
res->error_detail = "Help! it's again.";
if there's nothing to format, without mandating a superfluous pstrdup().)
v4 rebases, incorporates Chao Li's feedback, rewrites the surrounding
paragraph a bit, and fixes a silly misuse of <xref> when I meant
<link>.
--Jacob
v4-0001-oauth-Let-validators-provide-failure-DETAILs.patch
Description: Binary data
since-v3.nocfbot.diff
Description: Binary data
