Re: git: 97b0bb60dc4d - main - textproc/opensearch: Revert "Switch to a static pkg-plist"

On Mon, Sep 12, 2022 at 03:28:47PM -1000, Romain Tarti?re wrote:
> On Mon, Sep 12, 2022 at 04:27:13AM +0000, Alexey Dokuchaev wrote:
> > On Sun, Sep 11, 2022 at 09:02:39PM +0000, Romain Tarti??re wrote:
> > > commit 97b0bb60dc4dc6a71a26e08cc3d92576067b017a
> > > 
> > >   textproc/opensearch: Revert "Switch to a static pkg-plist"
> > > 
> > >   This reverts commit 43c14b71d3e7212d18e14b44911abffc99f40c9b.
> > 
> > Replying to a random backout commit with "this reverts commit $sha1"
> > commit log: are they being performed automatically?  Why they's never
> > a sensible commit message which would explain the reasons for revert?
> 
> Not sure I can help, but I can write the story of this commit:
> 
> I was unsure about the changes in
> 43c14b71d3e7212d18e14b44911abffc99f40c9b that's the reason why these
> changes where in a specific commit, and I noted the reason for this in
> the PR that suggested the change:
> https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=265704#c6

Okay, so it's about lots of embedded version numbers in *.jar files which
pollute the pkg-plist and you fear that subsequent updates would add some
extra churn you'd rather avoid.

> I also assume some tooling might exist and help you navigate easily in
> history, and they might be mislead by altered revert commit messages
> (the reason why I was not happy with amending the commit).

Navigating the history is not exactly what I'm concerned about. :-)

> So this is the reason _I_ keep the generated commit message and don't
> replace it with something more descriptive like "textproc/opensearch:
> Revert back to dynamic plist".

It is not more descriptive at all as it is still missing the particular
reason for the backout.  Regardless of your workflow and tangos you dance
with git, commit logs should not just describe *what* had been changed,
but more importantly, *why*.

./danfe