llvm dev - Jan 2015 - [LLVMdev] How to contact LLVM admins?

> We have had no trouble using "Patch by <name>" as the form
of attribution for a
> lot of years with contributions from a wide range of companies. So I'm
not really
> inclined to change a process which leaves the system of record *exactly*
the
>  same (a textual acknowledgement of attribution) and only makes one mirror
of
> it superficially better.
Adding --use-log-author to "git svn fetch" will have no effect unless
the committer either adds --add-author-from to "git svn dcommit" or
does a one-time setup:

    $ git config --global config svn.addAuthorFrom true

You can continue to use "Patch by <name>" if neither you nor the
author care.  If correct attribution on the git mirror is important to
*either* party, then the git mirror must be configured with
--use-log-author.  As a committer, correct attribution on the git
mirror is important to me for ethical and practical reasons.  It
should probably be important to me for legal reasons too.  :)

Thanks,
Greg

On Wed, Jan 21, 2015 at 3:10 PM, Chandler Carruth <chandlerc at
google.com> wrote:
>
> On Tue, Jan 20, 2015 at 10:15 PM, Greg Fitzgerald <garious at
gmail.com> wrote:
>>
>> Below is a copy of the email I sent to llvm-admin.  If it is lost in
>> the ether, can you reply on this thread?  There's an LLD
contributor
>> looking to add an ARMv7 backend.  It's a significant contribution
so
>> I'd like to land the patch under the correct author.
>>
>>
>> Hi LLVM admins,
>>
>> I'd like to commit patches on behalf of others, but I'm not
>> comfortable simply adding "Patch by <name>" to the log
message.  When
>> people use "git blame", they will see my name instead of the
original
>> author.
>>
>> We can fix this by adding the '--use-log-author' flag anywhere
the git
>> mirror currently invokes "git svn fetch".  This will cause
git-svn to
>> look for the text "From: <email>" in the commit message
and use it to
>> set the Author field in the git commit.  Committers can set the
"From:
>> <email>" text by hand or by using
'--add-author-from' when invoking
>> "git svn dcommit".
>>
>> Could someone please upgrade git-svn to a version supporting this flag
>> and update the git mirrors to call "git svn --use-log-author
fetch"?
>
>
> This wouldn't fix any of the historical commits made using the original
> pattern, and IMO is no better or worse than the current pattern.
>
> We have had no trouble using "Patch by <name>" as the form
of attribution
> for a lot of years with contributions from a wide range of companies. So
I'm
> not really inclined to change a process which leaves the system of record
> *exactly* the same (a textual acknowledgement of attribution) and only
makes
> one mirror of it superficially better.

On Wed, Jan 21, 2015 at 11:43 PM, Greg Fitzgerald <garious at gmail.com>
wrote:
> > We have had no trouble using "Patch by <name>" as the
form of
> attribution for a
> > lot of years with contributions from a wide range of companies. So
I'm
> not really
> > inclined to change a process which leaves the system of record
*exactly*
> the
> >  same (a textual acknowledgement of attribution) and only makes one
> mirror of
> > it superficially better.
>
> Adding --use-log-author to "git svn fetch" will have no effect
unless
> the committer either adds --add-author-from to "git svn dcommit"
or
> does a one-time setup:
>
>     $ git config --global config svn.addAuthorFrom true
>
> You can continue to use "Patch by <name>" if neither you
nor the
> author care.  If correct attribution on the git mirror is important to
> *either* party, then the git mirror must be configured with
> --use-log-author.  As a committer, correct attribution on the git
> mirror is important to me for ethical and practical reasons.  It
> should probably be important to me for legal reasons too.  :)

Not having different attribution mechanisms or creating any appearance that
one might be more or less important than another is important to me.
-------------- next part --------------
An HTML attachment was scrubbed...
URL:
<http://lists.llvm.org/pipermail/llvm-dev/attachments/20150121/c5fac750/attachment.html>

On 22 January 2015 at 07:50, Chandler Carruth <chandlerc at google.com>
wrote:
> Not having different attribution mechanisms or creating any appearance that
> one might be more or less important than another is important to me.
I think the worst part is that it'll give people the confidence that
git-blame is correct, and get people to believe that previous commits
(with "Patch by" comments) were actually made by the committers. This
would in turn, make me check the commit log anyway, which defies the
purpose.

I think I agree with Chandler that the benefits are not enough to
warrant such a change.

cheers,
--renato