PSC Vote: Keeping old NEWS in the news file for non-EOL versions

[Greg Troxel]

"Regina Obe" <lr at pcorp.us> writes:

> It has come to my attention that at least Paul and Sandro are inconvenienced
> by not having old NEWS in the news file.
> I had assumed finding old stuff using git log grep would be sufficient and
> keep our NEWS file not so long.

The longstanding GNU conventions are:
  - to have only NEWS in the NEWS file, not lists of bugfixes
  - to keep prepending NEWS for releases

> That said, how about this as a compromise.  Moving forward we'll keep all
> that crap in the NEWs file, but as soon as we call EOL on a version, we
> remove all remnants of it from the master NEWS.

There's no need for perjorative language; this is merely a mild
difference of opinion about the value of easily-accessible history vs
the perception of cleanliness.

Why do lines in a file, that is organized recent first, and which anyone
can choose to stop reading at any point, bother people that don't (this
minute) want to read them?

> The thing that annoyed me most about the NEWS file is that it had history
> going back to PostGIS 0.1.  So it was very very long. I assume only having
> non-EOLd versions wouldn't be too bad and would allow for easy scan of
> what's changed.

I don't follow easy scan.  You start at the top and it has NEWS for the
current release.  When you get to the previous release, you stop
reading, if that's what you want to do.    Or, e.g., if you're updating from
3.3 to 3.6 (because you never updated to 3.4 and 3.5), then you keep
reading until you get to 3.3 and then stop (before reading the 3.3
section).

I really don't understand why that's inconvenient for anyone.  Unless
they print the whole file to read on paper, but that seems very 90s.