Re: Is it time to drop ChangeLogs?

[Eli Zaretskii]

> From: Ingo Lohmar <address@hidden>
> Cc: address@hidden
> Date: Tue, 08 Mar 2016 22:25:35 +0100
> 
> On Tue, Mar 08 2016 22:37 (+0200), Eli Zaretskii wrote:
> > Using Git is not a problem for me.  The problem is that the
> > information in Git log is unreliable.  The other problem is that will
> > never succeed in teaching new contributors how to make good log
> > messages unless we have an easy way of fixing mistakes there.
> 
> 
> Some arguments in this thread are repeated ad infinitum although they
> don't seem to stand a little scrutiny.

That's because some people don't seem to read the thread, and keep
coming up with the same incorrect arguments time and again.

> "git log" messages cannot technically be both immutable and
> unreliable: At least there is some severely imprecise use of
> language going on.

You need to read the thread to understand what is being alluded to.
"Unreliable" in the sense that its text includes mistakes and cannot
be trusted.

> In contrast to your opinion, it seems to me that fixing mistakes in the
> Changelogs teaches a contributor who has committed with a flawed commit
> message that it's not really important.  They, or somebody else, can
> clean up their (incl. possibly my) mess.  As Oscar has argued, having
> the original commit rejected (by means to be discussed, and only until
> people have shown good judgment and discipline) teaches them that commit
> messages matter.

Emacs development doesn't work by requiring each commit be posted for
review as prerequisite for committing, so what Oscar suggests is not
possible.  (Please don't ask why, it was explained many times
already.)

> The whole argument for Changelogs comes down to a) being an established
> band-aid to clean up spilt milk, or b) providing a fixed-form summary of
> things that can be obtained using the VCS (provided the humans or tools
> wirting the Changelog are as "reliable" as the VCS).

Find a better and more reliable way of dealing with the problems
described here, and I'll be the first to agree not to reintroduce
ChangeLogs.