git/list[1] front-page[2] threads[3] people[4] search[5] about
 

Re: rebase flattens history when it shouldn't?

From
Sergey Organov <sorganov@gmail.com>
Date
Aug 6, 2014, 11:36 UTC
Message-ID
<87ppgdu9xo.fsf@osv.gnss.ru>
In-Reply-To
<20140723175218.GB12427@google.com>
Jonathan Nieder <jrnieder@gmail.com> writes:
Show 30 quoted lines
> Hi Sergei,
>
> Sergei Organov wrote:
>
>>      --C--
>>     /     \
>>    /   ----M topic,HEAD
>>   /   /
>>  A---B master
>>
>> shouldn't
>>
>> $ git rebase master
>>
>> be a no-op here?
> [...]
>> I'd expect --force-rebase to be required for this to happen:
>>
>> -f, --force-rebase
>>     Force the rebase even if the current branch is a descendant of the
>>     commit you are rebasing onto. Normally non-interactive rebase will
>>     exit with the message "Current branch is up to date" in such a
>>     situation.
> [...]
>> Do you think it's worth fixing?
>
> Thanks for a clear report.
>
> After a successful 'git rebase master', the current branch is always a
> linear string of patches on top of 'master'.
Is this documented? Except implicitly by the: 
-p, --preserve-merges
           Instead of ignoring merges, try to recreate them.
??
Anyway, why such a requirement, and is it actually enforced by tests?
> The "already up to date" behavior when -f is not passed is in a
> certain sense an optimization --- it is about git noticing that 'git
> rebase' wouldn't have anything to do (except for touching timestamps)
> and therefore doing nothing.

Maybe, but I'd argue it's rather sane behavior to do no rebase when new rebase point is the same as original in general. I.e., when "current branch is a descendant of the commit you are rebasing onto", as documentation says.

> So I don't think requiring -f for this case would be an improvement.

I still do, as it will match documentation, that in turn looks reasonable.

> I do agree that the documentation is misleading.

If the problem is in documentation, it's not only misleading, it's formally wrong.

> Any ideas for wording that could make it clearer?
Well, if it's indeed documentation, how about this:
diff --git a/Documentation/git-rebase.txt b/Documentation/git-rebase.txt
index 2a93c64..62dac31 100644
--- a/Documentation/git-rebase.txt
+++ b/Documentation/git-rebase.txt
@@ -316,10 +316,9 @@ which makes little sense.
 
 -f::
 --force-rebase::
-	Force the rebase even if the current branch is a descendant
-	of the commit you are rebasing onto.  Normally non-interactive rebase will
-	exit with the message "Current branch is up to date" in such a
-	situation.
+	Force the rebase even if the result will only change commit
+	timestamps. Normally non-interactive rebase will exit with the
+	message "Current branch is up to date" in such a situation.
 	Incompatible with the --interactive option.
 +
 You may find this (or --no-ff with an interactive rebase) helpful after

BTW, why "Incompatible with the --interactive option."? Isn't "force"
assumed by --interactive, functionally?
-- 
Sergey.
Previous: Sergey Organov
Message 6 of 6 in “rebase flattens history when it shouldn't?”
  1. Sergei OrganovJul 23, 2014
  2. Jonathan NiederJul 23, 2014
  3. Sergei OrganovJul 23, 2014
  4. Holger HellmuthAug 6, 2014
  5. Sergey OrganovAug 6, 2014
  6. Sergey OrganovAug 6, 2014

Read the whole thread, see it on lore, or plain text.

$ cat FOOTERMessages come from the public archive at lore.kernel.org/git, fetched every hour. The front page is chosen and written each morning by an AI editor and can be wrong; the threads themselves are the record. About and API. For agents: an MCP server at https://gitlist.dev/mcp, and any thread, story or person page as Markdown by adding .md to its URL (or sending Accept: text/markdown). Details in /llms.txt.