Re: [TCLCORE] Trunk seems forked in Tk

[Donal K. F. <don...@ma...>]

On 06/08/2012 21:11, Jan Nijtmans wrote:
> It's not even necessary to rename it, merging it (as DGP
> did) is sufficient. People updating to "trunk" will update to
> the merged node, because that's the one with the latest
> time-stamp. Note however that a simple "fossil update"
> might not be sufficient: That will only bring you to the
> latest successor of the current node, which might
> be the fork.

I never liked relying on that. Now that we know there's been a fork,
it's better to make one of the branches be non-canonical so that if
there are future commits against it, they *won't* be inclined to make
others doing a plain 'fossil update' jump between branches.

> The problem I see with renaming is that now Kevins work
> is marked as a "mistake", but it was not a mistake at
> all. It was just a temporary fork, that's all. Nothing wrong
> with that, as long as someone eventually resolves the
> fork. We're just not used to that, but its a
> normal phenomena in fossil.

It's a small mistake in that he didn't pull/update before committing the
patch. For a quiet project, that wouldn't be a big deal. For Tcl, it is
quite noticeable (we get quite a few commits per day); Tk is more
variable, sometimes quiet and sometimes not. The time spent discussing
this is far longer than the fixing took.

I've done this mistake too in the past. It's no big deal. :-)

Donal.