Freeze or no freeze?

[Martin von Zweigbergk]

The code freeze just started. I'd like to start a discussion about no
longer having freezes.

Some benefits of not having a freeze:

* Patch authors won't have to wait so long for feedback and won't risk
going as far in the wrong direction
* Review load gets spread out a little more evenly
* Those who don't particular care about running only released versions
won't have to wait so long for new features and bug fixes to land on the
default branch

I'm personally affected by all those points, at least to some degree. I'm
mentioning that just so my bias is clear.

As I understand it, the reason we have the freeze is in order to convince
people to test the release candidate and not a later version from the
default branch. I believe it's effective in achieving that (I mean that I
think most people do not instead run their own patched version). However,
I'm not sure we would find that many fewer bugs if we queued patches for
the default branch while the release candidate is out. It obviously happens
that a bug from the stable branch gets fixed by patch to the default branch
(and people testing on the default branch would thus not notice the bug on
the stable branch), but my impression is that that is pretty rare.

Thoughts? Were there other reasons for the code freeze that I missed?
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.mercurial-scm.org/pipermail/mercurial-devel/attachments/20180724/a0be5064/attachment.html>