[BusyBox] Re: Changelog

[Pavel Roskin]

Hello, Erik!

> > Shouldn't we switch to the GNU-style ChangeLog and NEWS immediately?
> > ChangeLog documents who did what and when in the chronological order.
> > 
> > NEWS documents only user visible the changes (well, reducing the size by
> > 3k _is_ a user visible change in the context of BusyBox).
> 
> I'm familier with GNU-style ChangeLogs (I used one for example on GnomeHack).
> But since this can be autogenerated (rcs2log > Changelog) it rarely got updated

Actually, there is another utility for that:
http://www.red-bean.com/~kfogel/cvs2cl.shtml

> and was not very useful for me.  Running 'cvs log' or visiting
> http://busybox.net/cgi-bin/cvsweb/ works just as well and is always up
> to date.

Both require that you are online.
Not useful with releases that one gets by ftp.

> Could you describe in more detail why you think this change is needed?  Is the
> current Changelog lacking in some way?  The current Changelog is intended to be
> descriptive of the user visible changes so that poeple don't need to check all
> the cvs logs and diffs.  I think the current Changelog does this quite well.  

First of all, don't forget about NEWS. Actually, the existing Changelog
can just be renamed to NEWS.

This file documents only user visible changes. Nothing else. Users should
(= are supposed to) use officially released version. When they upgrade
from one version to another they read NEWS to find out what has changed.

The users are not interested in bugs that have never existed in any
released version. Moreover, users don't normally want to know who made a
change.

On another side ChangeLog (note capital "L") exists for logging changes
(as it's name suggests). I don't know about other people, but it helps me
produce better patches if I document them. I have an additional chance to
review my changes.

ChangeLog is more formal than the CVS comments. It maybe an annoyance to
write it every time, but is is of great help if you are hunting a bug.

More motivation can be found at http://www.gnu.org/prep/standards_toc.html

My last argument is that doing things in a standard way is a good thing
unless there are reasons to do otherwise.

Pavel