Re: [PATCH] push: add config option to --force-with-lease by default.

Francesco Mazzoli [off-list ref] writes:

Moreover, it seems to me that the problem `--force-with-lease` is
just one of marketing. `--force-with-lease` is strictly more "safe"
than `--force` in the sense that it'll reject some pushes that `--force`
will let through.

By that logic, a hypothetical update to `--force` that makes 1/3 of
the attempted forced push randomly would make it safer than the
current `--force`, wouldn't it?

When third-party tools fetch and update remote-tracking branches
behind the users' back, the safety based on the stability of
remote-tracking branches are defeated.  And the biggest problem
is that the way `--force-with-lease` misbehaves---it is not like
it randomly and mistakenly stops the push that could go through;
it lets through what shouldn't.

See the other patch I sent just now---with something like that patch
that lets those like you, who know their remote-tracking branches
are reliable, use the lazy form, while disabling it by default for
others (until they examine their situation and perhaps disable the
problematic auto-fetching) in place, I do not think it is a bad idea
to advertise --force-with-lease a safer option than --force (because
those for whom it is not safer will not be able to use it).

On 6 July 2017 at 21:13, Junio C Hamano [off-list ref] wrote:

By that logic, a hypothetical update to `--force` that makes 1/3 of
the attempted forced push randomly would make it safer than the
current `--force`, wouldn't it?

It would. However, this additional safety is not really meaningful to any
workflow, while the one added by `--force-with-lease` is.

When third-party tools fetch and update remote-tracking branches
behind the users' back, the safety based on the stability of
remote-tracking branches are defeated.  And the biggest problem
is that the way `--force-with-lease` misbehaves---it is not like
it randomly and mistakenly stops the push that could go through;
it lets through what shouldn't.

See the other patch I sent just now---with something like that patch
that lets those like you, who know their remote-tracking branches
are reliable, use the lazy form, while disabling it by default for
others (until they examine their situation and perhaps disable the
problematic auto-fetching) in place, I do not think it is a bad idea
to advertise --force-with-lease a safer option than --force (because
those for whom it is not safer will not be able to use it).

Fair enough, I'm OK with enabling it with some config. I'd still like a
way to enable it by default if I want though.

Thanks,
Francesco