Tom Lane writes:
> > Well, a "release candidate" is "no more changes unless it's broken".
>
> True for code, but we have not in the past applied that standard to
> docs.
In the past, the docs were frozen two weeks before the release candidate
so that Thomas Lockhart could prepare them for publication. This time
we've been more relaxed, but I still think we have to stop somewhere. In
particular, a few days of rest is needed even for the docs, because:
* To make sure the documents are built at least once in full so they are
picked up when the release is built.
* To make sure the documents still build on whatever toolchain
postgresql.org is running. (There are occasional bugs in processors and
stylesheets.)
* To make sure everything looks OK when processed with whatever stylesheet
version postgresql.org uses.
* To make sure everything is spell-checked and edited for mistakes.
* To build the final edition of the man pages.
* To build printed versions and avoid post-release editing because of
certain "web-only" constructs or other things that don't work in the print
releases.
Some things to think about. If you think your changes don't affect these
issues, go ahead, but "overhaul" leaves me wondering.
--
Peter Eisentraut peter_e@gmx.net