Revert "[POWERPC] <various>"

Revert "[POWERPC] <various>"

[Simon Arlott]

There are a few powerpc patches plus reverts in the recent push to torvalds/linux-2.6 - shouldn't this be avoided whenever possible because it could cause problems for people trying to use git bisect?

-- 
Simon Arlott

Re: Revert "[POWERPC] <various>"

[Olaf Hering]

On Mon, Apr 30, Simon Arlott wrote:

> There are a few powerpc patches plus reverts in the recent push to 
> torvalds/linux-2.6 - shouldn't this be avoided whenever possible because it 
> could cause problems for people trying to use git bisect?

The reverts do not cause compile failures (I hope).
You may lose IDE if you happen to debug something on Pegasos.
I'm sure the real bisect breakers are elsewhere.

Re: Revert "[POWERPC] <various>"

[Olaf Hering]

On Mon, Apr 30, Simon Arlott wrote:

> There are a few powerpc patches plus reverts in the recent push to 
> torvalds/linux-2.6 - shouldn't this be avoided whenever possible because it 
> could cause problems for people trying to use git bisect?

The reverts do not cause compile failures (I hope).
You may lose IDE if you happen to debug something on Pegasos.
I'm sure the real bisect breakers are elsewhere.

Re: Revert "[POWERPC] <various>"

[Simon Arlott]

On 30/04/07 20:49, Olaf Hering wrote:
> On Mon, Apr 30, Simon Arlott wrote:
> 
>> There are a few powerpc patches plus reverts in the recent push to 
>> torvalds/linux-2.6 - shouldn't this be avoided whenever possible because it 
>> could cause problems for people trying to use git bisect?
> 
> The reverts do not cause compile failures (I hope).
> You may lose IDE if you happen to debug something on Pegasos.
> I'm sure the real bisect breakers are elsewhere.

The reverts themselves are not the real problem, a git bisect could occur 
between the commit adding it and the one that reverts it. If that commit 
introduced a bug then surely it would be better to avoid releasing it 
elsewhere even if it will be reverted in the same batch.

Of course, this is probably difficult to do with git and impossible if 
someone has cloned the bad commit already and tries to pull :/

Is it possible to mass copy all the good commits to create a clean branch?

-- 
Simon Arlott

Re: Revert "[POWERPC] <various>"

[Simon Arlott]

On 30/04/07 20:49, Olaf Hering wrote:
> On Mon, Apr 30, Simon Arlott wrote:
> 
>> There are a few powerpc patches plus reverts in the recent push to 
>> torvalds/linux-2.6 - shouldn't this be avoided whenever possible because it 
>> could cause problems for people trying to use git bisect?
> 
> The reverts do not cause compile failures (I hope).
> You may lose IDE if you happen to debug something on Pegasos.
> I'm sure the real bisect breakers are elsewhere.

The reverts themselves are not the real problem, a git bisect could occur 
between the commit adding it and the one that reverts it. If that commit 
introduced a bug then surely it would be better to avoid releasing it 
elsewhere even if it will be reverted in the same batch.

Of course, this is probably difficult to do with git and impossible if 
someone has cloned the bad commit already and tries to pull :/

Is it possible to mass copy all the good commits to create a clean branch?

-- 
Simon Arlott

Re: Revert "[POWERPC] <various>"

[Stephen Rothwell]

[-- Attachment #1: Type: text/plain, Size: 1026 bytes --]

On Mon, 30 Apr 2007 20:56:05 +0100 Simon Arlott <simon@fire.lp0.eu> wrote:
>
> The reverts themselves are not the real problem, a git bisect could occur
> between the commit adding it and the one that reverts it. If that commit
> introduced a bug then surely it would be better to avoid releasing it
> elsewhere even if it will be reverted in the same batch.

At least one of the reverts was just because the original commit clashed
with a commit in a different tree (netdev) so that will not cause and
build or runtime problems.

> Of course, this is probably difficult to do with git and impossible if
> someone has cloned the bad commit already and tries to pull :/

Not impossible, but it rebases the tree which means that people who have
pulled the tree need to force their next updates.

> Is it possible to mass copy all the good commits to create a clean branch?

It is, but in this case was probably not worth it.

--
Cheers,
Stephen Rothwell                    sfr@canb.auug.org.au
http://www.canb.auug.org.au/~sfr/

[-- Attachment #2: Type: application/pgp-signature, Size: 189 bytes --]

Re: Revert "[POWERPC] <various>"

[Stephen Rothwell]

[-- Attachment #1: Type: text/plain, Size: 1026 bytes --]

On Mon, 30 Apr 2007 20:56:05 +0100 Simon Arlott <simon@fire.lp0.eu> wrote:
>
> The reverts themselves are not the real problem, a git bisect could occur
> between the commit adding it and the one that reverts it. If that commit
> introduced a bug then surely it would be better to avoid releasing it
> elsewhere even if it will be reverted in the same batch.

At least one of the reverts was just because the original commit clashed
with a commit in a different tree (netdev) so that will not cause and
build or runtime problems.

> Of course, this is probably difficult to do with git and impossible if
> someone has cloned the bad commit already and tries to pull :/

Not impossible, but it rebases the tree which means that people who have
pulled the tree need to force their next updates.

> Is it possible to mass copy all the good commits to create a clean branch?

It is, but in this case was probably not worth it.

--
Cheers,
Stephen Rothwell                    sfr@canb.auug.org.au
http://www.canb.auug.org.au/~sfr/

[-- Attachment #2: Type: application/pgp-signature, Size: 189 bytes --]