Re: Fw: [PATCH 31/49] PCI: PCI/Cardbus cards hidden, needs pci=assign-busses to fix

Re: Fw: [PATCH 31/49] PCI: PCI/Cardbus cards hidden, needs pci=assign-busses to fix

[Greg KH]

On Thu, Mar 23, 2006 at 04:38:44PM -0800, Andrew Morton wrote:
> 
> (added Junio)

Added the git mailing list as everyone should get involved :)

> recap: the patch emails which Greg sends to linux-kernel do not identify
> their actual Author.
> 
> 
> Greg KH <greg@kroah.com> wrote:
> >
> > On Thu, Mar 23, 2006 at 04:15:21PM -0800, Andrew Morton wrote:
> > > 
> > > It's unclear from this email who the patch author was?
> > 
> > Git seems to strip that off when it converts them to emails.  It was:
> > 	From: Bernhard Kaindl <bk@suse.de>
> > 
> > If you look at the git changeset, it got it correct.
> > 
> 
> OK.   We really should have a
> 
> 	From: Bernhard Kaindl <bk@suse.de>
> 
> right at the start of the email, IMO.  Can you describe how you're producing
> them please?

I'm using:
	git format-patch -n origin..HEAD
to generate the raw patch files, and then:
	git-send-email --in-reply-to "<some_message_id>" --to some_mailing_list@somewhere.com

fixing the obvious message id and mailing list address to be the correct
one depending on the subsystem the patches are from.

thanks,

greg k-h

Re: Fw: [PATCH 31/49] PCI: PCI/Cardbus cards hidden, needs pci=assign-busses to fix

[Greg KH]

On Thu, Mar 23, 2006 at 04:46:54PM -0800, Greg KH wrote:
> On Thu, Mar 23, 2006 at 04:38:44PM -0800, Andrew Morton wrote:
> > 
> > (added Junio)
> 
> Added the git mailing list as everyone should get involved :)
> 
> > recap: the patch emails which Greg sends to linux-kernel do not identify
> > their actual Author.
> > 
> > 
> > Greg KH <greg@kroah.com> wrote:
> > >
> > > On Thu, Mar 23, 2006 at 04:15:21PM -0800, Andrew Morton wrote:
> > > > 
> > > > It's unclear from this email who the patch author was?
> > > 
> > > Git seems to strip that off when it converts them to emails.  It was:
> > > 	From: Bernhard Kaindl <bk@suse.de>
> > > 
> > > If you look at the git changeset, it got it correct.
> > > 
> > 
> > OK.   We really should have a
> > 
> > 	From: Bernhard Kaindl <bk@suse.de>
> > 
> > right at the start of the email, IMO.  Can you describe how you're producing
> > them please?
> 
> I'm using:
> 	git format-patch -n origin..HEAD
> to generate the raw patch files, and then:
> 	git-send-email --in-reply-to "<some_message_id>" --to some_mailing_list@somewhere.com

Oops, forgot the "*.txt" at the end of that last line to specify the
patches to send out.

thanks,

greg k-h

Re: Fw: [PATCH 31/49] PCI: PCI/Cardbus cards hidden, needs pci=assign-busses to fix

[Junio C Hamano]

Greg KH <greg@kroah.com> writes:

> I'm using:
> 	git format-patch -n origin..HEAD
> to generate the raw patch files, and then:
> 	git-send-email --in-reply-to "<some_message_id>" --to some_mailing_list@somewhere.com
>
> fixing the obvious message id and mailing list address to be the correct
> one depending on the subsystem the patches are from.

I think format-patch does the right thing (I wrote it), but I am
not sure what send-email does wrt the From: header.  Who wrote
the send-email anyway?  I see your name on it ;-)

The cleanest way send-email should handle a patch authored by
somebody other than you, I think, is to still use From: to name
the author (format-patch output records the author on From:
line), and use Sender: of the outgoing e-mail to record that the
message is from you.  I suspect it probably doesn't.

The second best would be to add the duplicated From: to name the
author (who is _not_ you) to the top of the body of the message.
I do not particularly like that format myself, though.  Sender:
header was invented to send an e-mail authored by somebody other
than the sender of the message at the mail transport level, long
before Documentation/SubmittingPatches were written and git was
invented, and somehow I think that is a more kosher way to
handle that than the "extra From: at the beginning of the
message" clutch recommended in SubmittingPatches document.  

On the acceptance side, "git am" (or "git applymbox") should be
able to handle either format.

Re: Fw: [PATCH 31/49] PCI: PCI/Cardbus cards hidden, needs pci=assign-busses to fix

[Andrew Morton]

Junio C Hamano <junkio@cox.net> wrote:
>
> The second best would be to add the duplicated From: to name the
> author (who is _not_ you) to the top of the body of the message.
> I do not particularly like that format myself, though.  Sender:
> header was invented to send an e-mail authored by somebody other
> than the sender of the message at the mail transport level, long
> before Documentation/SubmittingPatches were written and git was
> invented, and somehow I think that is a more kosher way to
> handle that than the "extra From: at the beginning of the
> message" clutch recommended in SubmittingPatches document.  

The email I received from Greg had no Sender: header at all.  I could find
no indication of who authored the patch in that email.

The convention of adding the From: to the top of the body of the changelog
is explicit and simple - I think it's a reasonable thing to do.

We wouldn't want to attempt to mix this concept up with email envelopes or
email headers or anything like that.  The authorship is an attribute of the
patch, and has nothing to do with how it was transported, stored or
anything like that.

Re: Fw: [PATCH 31/49] PCI: PCI/Cardbus cards hidden, needs pci=assign-busses to fix

[Junio C Hamano]

Andrew Morton <akpm@osdl.org> writes:

> We wouldn't want to attempt to mix this concept up with email envelopes or
> email headers or anything like that.  The authorship is an attribute of the
> patch, and has nothing to do with how it was transported, stored or
> anything like that.

Fair enough.  This is the approach I called "the second best" in
my message but I am inclined to agree with you.

This was tested once by sending myself two patches.

-- >8 --
[PATCH] send-email: Identify author at the top when sending e-mail

git-send-email was not checking if the sender is the same as the
patch author.  Follow the "From: at the beginning" convention to
propagate the patch author correctly.

Signed-off-by: Junio C Hamano <junkio@cox.net>
---
diff --git a/git-send-email.perl b/git-send-email.perl
index 7c8d512..b220d11 100755
--- a/git-send-email.perl
+++ b/git-send-email.perl
@@ -307,6 +307,7 @@ $subject = $initial_subject;
 foreach my $t (@files) {
 	open(F,"<",$t) or die "can't open file $t";
 
+	my $author_not_sender = undef;
 	@cc = @initial_cc;
 	my $found_mbox = 0;
 	my $header_done = 0;
@@ -321,7 +322,12 @@ foreach my $t (@files) {
 					$subject = $1;
 
 				} elsif (/^(Cc|From):\s+(.*)$/) {
-					next if ($2 eq $from && $suppress_from);
+					if ($2 eq $from) {
+						next if ($suppress_from);
+					}
+					else {
+						$author_not_sender = $2;
+					}
 					printf("(mbox) Adding cc: %s from line '%s'\n",
 						$2, $_) unless $quiet;
 					push @cc, $2;
@@ -360,6 +366,9 @@ foreach my $t (@files) {
 		}
 	}
 	close F;
+	if (defined $author_not_sender) {
+		$message = "From: $author_not_sender\n\n$message";
+	}
 
 	$cc = join(", ", unique_email_list(@cc));
 

Re: Fw: [PATCH 31/49] PCI: PCI/Cardbus cards hidden, needs pci=assign-busses to fix

[Greg KH]

On Thu, Mar 23, 2006 at 06:27:15PM -0800, Junio C Hamano wrote:
> Andrew Morton <akpm@osdl.org> writes:
> 
> > We wouldn't want to attempt to mix this concept up with email envelopes or
> > email headers or anything like that.  The authorship is an attribute of the
> > patch, and has nothing to do with how it was transported, stored or
> > anything like that.
> 
> Fair enough.  This is the approach I called "the second best" in
> my message but I am inclined to agree with you.
> 
> This was tested once by sending myself two patches.

Oops, just saw this after I sent out the last set of patches.  It looks
good to me, I'll try it out next time.

And yes, I did write the original version of this perl script, but it's
been fixed up and made useful by Ryan.

thanks,

greg k-h

Re: Fw: [PATCH 31/49] PCI: PCI/Cardbus cards hidden, needs pci=assign-busses to fix

[Junio C Hamano]

Greg KH <greg@kroah.com> writes:

>> This was tested once by sending myself two patches.
>
> Oops, just saw this after I sent out the last set of patches.  It looks
> good to me, I'll try it out next time.

Thanks.  Will place this in "master" tonight.

Re: Fw: [PATCH 31/49] PCI: PCI/Cardbus cards hidden, needs pci=assign-busses to fix

[Ryan Anderson]

On Thu, Mar 23, 2006 at 06:27:15PM -0800, Junio C Hamano wrote:
> [PATCH] send-email: Identify author at the top when sending e-mail
> 
> git-send-email was not checking if the sender is the same as the
> patch author.  Follow the "From: at the beginning" convention to
> propagate the patch author correctly.

This looks fine.
My initial thought here was that it might be even simpler to just always
put the From: line in the body, with the patch author's name there, but
this is a bit cleaner, so ACK.

-- 

Ryan Anderson
  sometimes Pug Majere