Re: BK kernel commits list

Re: BK kernel commits list

[Hank Leininger]

On 2002-10-09, "David S. Miller" <davem@redhat.com> wrote: 
 
> I've created: 
>     bk-commits-head@vger.kernel.org 
>     bk-commits-24@vger.kernel.org 
 
FYI, I added these to the MARC list archives: 
 
http://marc.theaimsgroup.com/?l=bk-commits-head&r=1&w=2 
http://marc.theaimsgroup.com/?l=bk-commits-24&r=1&w=2 
 
...Sometime last night, so they've got all of today's traffic. 
 
-- 
Hank Leininger <hlein@progressive-comp.com>  
   

Re: New BK License Problem?

[Jeff Garzik]

Skip Ford wrote:
> What I would like to see is a 'linux-commits' mailing list where
> Woodhouse's per changeset diffs can all be posted.

Not a bad idea...  Various people have called for a patches mailing list 
in the past.


> I appreciate that he does it.  It's a nice service he provides.  But I'm
> sure Linus could write a script that mails 'linux-commits' with a diff
> for each changeset easily.

I'm sure Linus could write and maintain my net drivers for me, too.  :)

Maybe you could submit a modification to dwmw2's script to him, or run 
this yourself...?

	Jeff

Re: New BK License Problem?

[David Woodhouse]

skip.ford@verizon.net said:
>  I sort of had vger in mind, but I could set up a crude read-only list
> of some sort if need be on my dynamic IP line.

If a list is set up on vger I'll feed the patches to it.

>  I can't seem to find dwmw2's script.. 

CVSROOT=:pserver:anoncvs@cvs.infradead.org:/home/cvs
grep -q $CVSROOT ~/.cvspass || ( echo "$CVSROOT Ay=0=h<Z" >> ~/.cvspass )
cvs -d $CVSROOT co bkexport

--
dwmw2

Re: New BK License Problem?

[David Woodhouse]

davem@redhat.com said:
>  Should we have two lists, one for 2.4 and one for 2.5?
> I'll set it up once decided.

Probably one for each. We could add headers which say which branch it is 
but that's still a lot of extra traffic for subscribers who only want the 
stable branch info and will filter the 2.5 ones to /dev/null.

How about
 bk-commits-head
 bk-commits-2.4

then later bk-commits-2.6 etc...


--
dwmw2

BK kernel commits list

[David S. Miller]

I've created:

    bk-commits-head@vger.kernel.org
    bk-commits-24@vger.kernel.org

(note: no "period" between the 2 and the 4)

Have at it.

Re: BK kernel commits list

[Skip Ford]

David S. Miller wrote:
> 
> I've created:
> 
>     bk-commits-head@vger.kernel.org
>     bk-commits-24@vger.kernel.org
> 
> (note: no "period" between the 2 and the 4)
> Have at it.

Does this list need to allow posting as it does now?  Only one address
needs write permission, or maybe a few addresses as backup.  Discussion
should still happen on lkml.

-- 
Skip

Re: BK kernel commits list

[David S. Miller]

   From: Skip Ford <skip.ford@verizon.net>
   Date: Wed, 9 Oct 2002 07:49:32 -0400
   
   Does this list need to allow posting as it does now?  Only one address
   needs write permission, or maybe a few addresses as backup.  Discussion
   should still happen on lkml.

I can set this up once I know what the feeder's From addrss
looks like.

Re: BK kernel commits list

[David Woodhouse]

davem@redhat.com said:
>  I can set this up once I know what the feeder's From addrss looks
> like.

Well I was thinking of using `bk changes -d:USER:@:HOST:`

Can we make it accept mail only with an envelope sender of 
dwmw2@.*.kernel.org?

--
dwmw2

Re: BK kernel commits list

[David S. Miller]

   From: David Woodhouse <dwmw2@infradead.org>
   Date: Wed, 09 Oct 2002 13:17:58 +0100
   
   Can we make it accept mail only with an envelope sender of 
   dwmw2@.*.kernel.org?

Any particular reason why the ".*." part can't be constant?

Re: BK kernel commits list

[Jeff Garzik]

A couple of suggestions for the output...

The BitKeeper header is unfortunately less than useful at times -- often 
multiple cset descriptions wind up in the header -- so I would suggest 
exporting the patch with "-hdu" instead of "-du", and then manually 
adding the changeset info and description at the top.

Also, diffstat output prepended would be nice too.

Re: BK kernel commits list

[Ben Collins]

On Wed, Oct 09, 2002 at 10:11:55AM -0400, Jeff Garzik wrote:
> A couple of suggestions for the output...
> 
> The BitKeeper header is unfortunately less than useful at times -- often 
> multiple cset descriptions wind up in the header -- so I would suggest 
> exporting the patch with "-hdu" instead of "-du", and then manually 
> adding the changeset info and description at the top.
> 
> Also, diffstat output prepended would be nice too.

Just please make sure that the changeset info where it describes all the
files in the delta. I.e. the ones that are moved, deleted, new. There's
no way to deduce moves from the patch.

-- 
Debian     - http://www.debian.org/
Linux 1394 - http://www.linux1394.org/
Subversion - http://subversion.tigris.org/
Deqo       - http://www.deqo.com/

Re: BK kernel commits list

[David Woodhouse]

[-- Attachment #1: Type: text/plain, Size: 548 bytes --]


bcollins@debian.org said:
>  Just please make sure that the changeset info where it describes all
> the files in the delta. I.e. the ones that are moved, deleted, new.
> There's no way to deduce moves from the patch. 

This bit?

# This patch includes the following deltas:
#                  ChangeSet    1.713   -> 1.714
#       arch/i386/math-emu/poly.h       1.3     -> 1.4
#

Any idea how to get it other than 'bk export -tpatch | sed' ?

I need to do some real work... play with ths script and sort it out between 
yourselves :)


--
dwmw2


[-- Attachment #2: mailcset.sh --]
[-- Type: application/x-sh , Size: 858 bytes --]

Re: BK kernel commits list

[Ben Collins]

On Wed, Oct 09, 2002 at 03:55:00PM +0100, David Woodhouse wrote:
> 
> bcollins@debian.org said:
> >  Just please make sure that the changeset info where it describes all
> > the files in the delta. I.e. the ones that are moved, deleted, new.
> > There's no way to deduce moves from the patch. 
> 
> This bit?
> 
> # This patch includes the following deltas:
> #                  ChangeSet    1.713   -> 1.714
> #       arch/i386/math-emu/poly.h       1.3     -> 1.4
> #

Yeah, that's it.

> Any idea how to get it other than 'bk export -tpatch | sed' ?

Probably some sort of "bk changes" output. Try checking the -d spec
options.

> I need to do some real work... play with ths script and sort it out between 
> yourselves :)

I can't. Requires using BK :)

-- 
Debian     - http://www.debian.org/
Linux 1394 - http://www.linux1394.org/
Subversion - http://subversion.tigris.org/
Deqo       - http://www.deqo.com/

Re: BK kernel commits list

[Jeff Garzik]

Actually, after subscribing to bk-commits-* and seeing the output, I 
really think the only addition needed is to pipe the output through 
diffstat.

Re: BK kernel commits list

[Robert Love]

On Wed, 2002-10-09 at 15:48, Jeff Garzik wrote:

> Actually, after subscribing to bk-commits-* and seeing the output,
> I really think the only addition needed is to pipe the output
> through diffstat.

I think this would be cool, too.

I think simplicity is important, but more information is always helper.

I also like Daniel's suggestion of having the sender's _name_ by the
name attributed with the patch (or at least "BK Commit" or something).

But nothing _needs_ to be done.  This works great.  Good job, Dave and
Co.

	Robert Love

Re: BK kernel commits list

[Skip Ford]

Robert Love wrote:
> On Wed, 2002-10-09 at 15:48, Jeff Garzik wrote:
> 
> > Actually, after subscribing to bk-commits-* and seeing the output,
> > I really think the only addition needed is to pipe the output
> > through diffstat.
> 
> I think this would be cool, too.
> 
> I also like Daniel's suggestion of having the sender's _name_ by the
> name attributed with the patch (or at least "BK Commit" or something).
> 
> But nothing _needs_ to be done.  This works great.  Good job, Dave and
> Co.

I agree.  Thanks to David Woodhouse for feeding it and DaveM for
creating it.

I'd like to suggest that once Linus makes a release and all of the
changeset diffs are wiped out on kernel.org, that a message be sent to
bk-commits indicating that.  Nothing fancy, just an empty message with a
subject of "--MARK--" would be nice.  But either way, it's a great
list.

-- 
Skip

Re: BK kernel commits list

[Jeff Garzik]

Robert Love wrote:
> But nothing _needs_ to be done.  This works great.  Good job, Dave and
> Co.


Indeed: kudos and thanks.

Re: BK kernel commits list

[Ben Collins]

On Wed, Oct 09, 2002 at 03:48:58PM -0400, Jeff Garzik wrote:
> Actually, after subscribing to bk-commits-* and seeing the output, I 
> really think the only addition needed is to pipe the output through 
> diffstat.

Without the info concerning the file revs being affected, it's pretty
useless for me. I can actually just use that info and pull diffs
locally from my repo via sccs.

But I guess the point of the list is for diffs to kernel devs. Guess
I'll get my info elsewhere.


-- 
Debian     - http://www.debian.org/
Linux 1394 - http://www.linux1394.org/
Subversion - http://subversion.tigris.org/
Deqo       - http://www.deqo.com/

Re: BK kernel commits list

[David Woodhouse]

bcollins@debian.org said:
>  Without the info concerning the file revs being affected, it's pretty
> useless for me. I can actually just use that info and pull diffs
> locally from my repo via sccs.

We can do that. Worst case, someone needs to teach me enough sed to grok 
the 'bk export -tpatch' output, find the line matching
'^# This patch contains the following deltas:$' and spew output from that 
till the first line matching '^#$'. Ideally, there'll be an easier 
way of getting the info from bitkeeper though.

Oh, and would anyone mind if, after feeding the mail through diffstat and 
seeing zero lines of change, I aborted sending the mail?

Further criticism of the scripts will only be accepted in diff -u form 
unless you have an excuse at least as good as having been told personally 
by Larry that you're not allowed to use BitKeeper.

CVSROOT=:pserver:anoncvs@cvs.infradead.org:/home/cvs
grep -q $CVSROOT ~/.cvspass || ( echo "$CVSROOT Ay=0=h<Z" >> ~/.cvspass )
cvs -d $CVSROOT co bkexport

--
dwmw2

Re: BK kernel commits list

[Robert Love]

On Wed, 2002-10-09 at 16:52, David Woodhouse wrote:

> Oh, and would anyone mind if, after feeding the mail through diffstat and 
> seeing zero lines of change, I aborted sending the mail?

You mean so we did not see empty emails (usually those of the form
"merge from x into y")?

Not only do I not mind, I encourage... please do so.

	Robert Love

Re: BK kernel commits list

[David Woodhouse]

jgarzik@pobox.com said:
>  Actually, after subscribing to bk-commits-* and seeing the output, I
> really think the only addition needed is to pipe the output through
> diffstat.

Er, I did that last time you asked. Maybe it'll work better if I cvs update 
on hera after committing the change? :)


--
dwmw2