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

RE: pushing a delete-only commit consumes too much traffic

From
Randall S. Becker <rsbecker@nexbridge.com>
Date
Jan 26, 2018, 01:56 UTC
Message-ID
<000f01d39648$d918f6e0$8b4ae4a0$@nexbridge.com>
In-Reply-To
<79fd9c0b-3067-afce-1fcd-507b7c97889e@gmail.com>
So the point here is that the jars are still in the repository. They are deleted in your branch, but any objects depending on others (without a lot more information on your exact repository structure) may cause packed objects to be sent upstream. The delete is local to your branch, but the delete does not mean the objects are actually removed from your repository. To actually effect a removal, the objects would need to be no longer accessible, thus git gc would remove them permanently. That may be difficult depending on security on your upstream repository and what you are able to do there. So:
A--- B--- C
If A has no jars, B has the jars and has A as its parent, and C has no jars, but has B as its parent, then B is still accessible and the jars still exist in the repository but only not at the HEAD of your branch. Your tree may vary. How a push gets objects from your repository to your upstream depends on more information that I have but the point is that the jars still exist in a distributed sense. Your repository state and the upstream repository state do not need to be (and are likely not) identical.
Show 43 quoted lines
> -----Original Message-----
> From: Basin Ilya [mailto:basinilya@gmail.com]
> Sent: January 25, 2018 10:08 AM
> To: Randall S. Becker <rsbecker@nexbridge.com>; git@vger.kernel.org
> Subject: Re: pushing a delete-only commit consumes too much traffic
> 
> > Were the 60Mb of jars previously pushed in a commit that already existed
> on the upstream?
> yes
> 
> > Was the delete an actual removal of history or did you commit with the jars
> deleted, then pushed?
> I committed with the jars deleted
> 
> > Did you do a merge squash or delete branch to effect the removal.
> No
> 
> 
> 
> 
> On 25.01.2018 17:24, Randall S. Becker wrote:
> > On January 25, 2018 9:15 AM, Basin Ilya wrote:
> >
> >> I had a 60Mb worth of unneeded jar files in the project. I created a
> >> new branch and performed `git rm` on them. Now while I was pushing
> >> the change the counter of sent data reached 80Mb. Why is that?
> >
> > Can you provide more info? Were the 60Mb of jars previously pushed in a
> commit that already existed on the upstream? Was the delete an actual
> removal of history or did you commit with the jars deleted, then pushed? Did
> you do a merge squash or delete branch to effect the removal. More info
> please.
> >
> > Cheers,
> > Randall
> >
> > -- Brief whoami:
> >   NonStop developer since approximately NonStop(211288444200000000)
> >   UNIX developer since approximately 421664400
> > -- In my real life, I talk too much.
> >
> >
> >
Previous: Basin Ilya
Message 4 of 4 in “pushing a delete-only commit consumes too much traffic”
  1. Basin IlyaJan 25, 2018
  2. Randall S. BeckerJan 25, 2018
  3. Basin IlyaJan 25, 2018
  4. Randall S. BeckerJan 26, 2018

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.