Re: Should we restore manually maintained ChangeLogs

[Paul Eggert]

On 03/09/2016 11:57 AM, Eli Zaretskii wrote:
> > Like it or not, that is a form of historical research. 
> No, it's a very far cry from historical research.

I won't insist on calling it "historical research", admittedly a phrase 
that suggests bigger things than finding out why Emacs doesn't use 
setenv on "TZ". Still, the point remains that primary sources are 
unreliable in Emacs development, just as they are in any form of 
historical research. If someone ever gets around to writing a definitive 
history of Emacs (I'm looking at you, ESR!), then they'll have to take 
ChangeLogs with a grain of salt, just as you and I do in routine code 
spelunking.

> That cost is much lower than any of the alternatives proposed so far, 
> including the current arrangement with ChangeLog.2. It worked for years.

I'm afraid we'll have to disagree on costs. The old way of doing things 
was a constant irritation to me and to others.

> > Regardless of the approach taken, there is also a cost to
> > sprucing up the historical record
> Since this is regardless of the approach, it shouldn't affect the
> decision in this matter.

No, they're still related. If sprucing up ChangeLogs is low-priority 
work that distracts us from other things, then it's not advantageous to 
adopt a technical approach merely on the grounds that the approach makes 
it easier to spruce up ChangeLogs.