Re: Using VC for change descriptions

[Paul Eggert]

On 01/29/2018 05:30 AM, Joseph Myers wrote:
> It's a different approach from looking at lists of entities for each
> commit, but I think it's just as powerful in practice.

In my experience that is an understatement. When I'm trying to debug or 
spelunk code, "git blame" is significantly more useful than looking at 
entity lists in ChangeLogs, even when the ChangeLogs are well-written 
(which often they aren't).

For example, this morning I used used "git blame" when cleaning up some 
cruft in GNU Emacs's Fcall_interactively function. In this cleanup I 
removed a workaround for a ~30-year-old compiler bug that is no longer 
relevant. "git blame" let me easily determine the age of the workaround. 
It would have taken me far more time to use ChangeLogs to do the same 
thing, so I didn't bother. (When composing this email I just now took 
the time to look through the Emacs ChangeLogs, and they did not contain 
the relevant information so it would have been a waste of time to look 
at the ChangeLogs anyway.)

This experience is typical for me, when I hack on GNU programs. I almost 
never use the entity information in ChangeLog entries, as it's far less 
useful than "git blame". I do refer to the motivating text in ChangeLog 
entries, as that provides info that is not obvious from "git blame".