Re: Git, where zombie branches shamble again

[G. Branden Robinson]

I had a longer response to this, but it's probably better to cut to the
chase.  First I'll quote the charge sheet.

At 2021-10-24T20:35:44+0100, Keith Marshall wrote:
> to push the first change in my local patch queue ... and incidentally,
> (and unexpectedly), restore the public repository history, which you
> had (IMO inadvisedly) rewritten.
[...]
> That seems like *really* antisocial behaviour!  We had a similar
> discussion on MinGW, several years ago, when Earnie Boyd *proposed*
> rewriting history on a public repository server; Chuck Wilson and I
> persuaded him not to do so, because it really screws up collaborative
> effort ... even git's commit documentation *strongly* discourages it!
[...]
> (inspection of the actual "hg outgoing" output, without the grep
> filtering, shows that these *do* correspond to your rewritten
> history):
[...]
> What you describe may be a manifest *effect* of the problem; the
> *real* problem is that, contrary to established wisdom, you've abused
> git's ability to rewrite history, within a public repository — that
> should be an absolute taboo.  The only way *I* (and I guess, other
> collaborators) can recover from the adverse consequences of such
> history rewriting is to abandon my current working copy of the
> repository, clone a fresh copy, and forward port my outstanding patch
> queue to the newly cloned working copy.
[...]
> I have now cloned a fresh working copy of the public repository, so it
> is unlikely that *I* will be pushing a resurrection of this particular
> unwanted branch, but unless everyone else does likewise, perhaps we
> shouldn't be too surprised if it reappears again.

At 2021-10-24T13:18:24-0700, Larry McVoy wrote:
> On Sun, Oct 24, 2021 at 08:35:44PM +0100, Keith Marshall wrote:
> > What you describe may be a manifest *effect* of the problem; the
> > *real* problem is that, contrary to established wisdom, you've
> > abused git's ability to rewrite history, within a public repository
> > ??? that should be an absolute taboo.  
> 
> +1 except in extraordinary circumstances.

Since I am now accused four times over of rewriting history, and
moreover of violating an "absolute taboo", I must insist upon the
presentation of particulars.

1. What history has been altered?  Be specific.
2. Is any valuable information now missing from Savannah?
3. If so, can that information be recovered?
4. Is the culprit (me) unwilling to participate, alone or in concert
   with others, in the restoration of that information?
5. Is the culprit (me) accused of acting with malice?
6. What is the correct procedure for marking a development branch
   closed, abandoned, or otherwise defunct?

If you're "merely" trying to whack me over the head hard enough that I
don't run the command "git push --delete origin $branch" on the groff
Git repository again, worry not: I have no such intention.

Beyond that, I lack interest in engaging you any further on this subject
until you address the above questions...or elect to escalate the issue.

Regards,
Branden

signature.asc
Description: PGP signature