Guidelines for huge spec changelogs?

[Nigel Jones]

Michael Schwendt wrote:
> Some [or many?] packages in the collection contain really huge spec
> %changelog sections, which contain entries that date back as far as 
> the middle 90's.
> 
> It is not unusual that entries, which are several years old, have become
> irrelevant or wrong or get lost among several dozen entries like "rebuilt"
> or "update to x.y.z". The changelogs however are included in the rpm
> packages completely and also in web pages generated from them, e.g. in
> koji or repoview.
> 
> Do we have any guidelines about what to do with them and whether it
> is permitted to strip the changelogs from time to time? (and if so,
> whether to archive them anywhere other than in cvs?)
> 
> Not a big issue, not at all, I'm just curious. ;o)
> 
I think it's useful to have 1-2 releases worth of entries.  i.e., we are
entering into the dev cycle for F8, so F7 changes + F8 changes included.

I don't know, how about a %{name}.changelog for archived entries,
distributed via CVS?

A good example of this is Anaconda, I loaded the spec file by accident
on a slow link, and I'm sure the change log is about 5 times the size of
the actual spec file bits and bobs.

N.J.