threads / discuss / 6384

Re: Commit signing

Subject: Re: Commit signing

## tl;dr

3 messages between Jan 15, 2007 and Jan 15, 2007.

replies: 2people: 3as markdown or json

Horst H. von Brand· Jan 15, 2007, 13:01 UTC · lore

Andy Parkins <andyparkins@gmail.com> wrote:

[...]

> Again true.  What has that to do with Git though?  Why shouldn't Git have 
> features that let people with different methods of development from you use 
> it?

As long as nobody else ends up paying the cost...

>     It is certainly true that signed commits /is/ a feature.  And it's a 
> feature that some people might want.  If there isn't a technical argument 
> against it, what does it matter?

It needs to be accomodated in the commit object format, so it means an(other) incompatible change there. Need to add checking for properly signed commits all the way when slurping in a stream of changes. Need to set the whole up so it can bail out as if nothing ever happened in case one commit doesn't check out (this is probably easy).

> (Note: it doesn't matter enough to me that I would put the time in, I'm 
> arguing in the abstract really - should features be kept out because they 
> allow a development method we would find distasteful?)

I'd vote for "If somebody needs this bad enough, they are free to fork pgit ('paranoid git')", and add "if the changes look sane, and don't screw up the base (too much), we'll talk about merging back".

Open source development, scratch your own itch, bazaar and all that.

-- 
Dr. Horst H. von Brand                   User #22616 counter.li.org
Departamento de Informatica                    Fono: +56 32 2654431
Universidad Tecnica Federico Santa Maria             +56 32 2654239
Casilla 110-V, Valparaiso, Chile               Fax:  +56 32 2797513

Johannes Schindelin· Jan 15, 2007, 13:45 UTC · re: Horst H. von Brand · lore

Hi,

On Mon, 15 Jan 2007, Horst H. von Brand wrote:

Show 8 quoted lines

> Andy Parkins <andyparkins@gmail.com> wrote:
> 
> >     It is certainly true that signed commits /is/ a feature.  And it's a 
> > feature that some people might want.  If there isn't a technical argument 
> > against it, what does it matter?
> 
> It needs to be accomodated in the commit object format, so it means
> an(other) incompatible change there.

No. A signature can be appended to any text (which a raw commit buffer is), and git allows for appending that signature before writing the object (which also gives it the final object name). So, no commit object format change necessary.

Ciao, Dscho

Nicolas Pitre· Jan 15, 2007, 16:47 UTC · re: Horst H. von Brand · lore

On Mon, 15 Jan 2007, Horst H. von Brand wrote:

Show 14 quoted lines

> Andy Parkins <andyparkins@gmail.com> wrote:
> 
> > Again true.  What has that to do with Git though?  Why shouldn't Git have 
> > features that let people with different methods of development from you use 
> > it?
> 
> As long as nobody else ends up paying the cost...
> 
> >     It is certainly true that signed commits /is/ a feature.  And it's a 
> > feature that some people might want.  If there isn't a technical argument 
> > against it, what does it matter?
> 
> It needs to be accomodated in the commit object format, so it means
> an(other) incompatible change there.

Not really. Nothing prevents from having a signature block at the bottom of the commit message, just like signed tags.

> Need to add checking for properly
> signed commits all the way when slurping in a stream of changes. Need to
> set the whole up so it can bail out as if nothing ever happened in case one
> commit doesn't check out (this is probably easy).

... and if you don't care about the signature you may ignore it entirely.

Nicolas