Re: update hook for git to check push commit message

[Khem Raj <raj.khem@gmail.com>]

On (28/01/10 21:52), Phil Blundell wrote:
> On Thu, 2010-01-28 at 12:00 -0800, Khem Raj wrote:
> > Attached is a small hook for updates that are pushed into repo.
> > Right now it only checks the first line of the commit and expects
> > module: summary
> 
> I'm not really in favour of applying the commit message policy quite
> that harshly.  The wiki page that you linked to doesn't really seem to
> support this kind of check: the actual guidance it gives about checkin
> messages is:

Thats true. However I have seen many devs complain about the module:
summary. That was my motivation. 

> 
> * Have a clear commit message (example):
>     - The first line of commit is a summary of the changes.
>     - The first line should start with the name of the recipe the change affects.
>     - The rest of the message should give more details on the change as appropriate.
>     - Mention the affected bug numbers if appropriate.
>     - Give credit where credit is due. If you commit someone else's work more or 
>       less verbatim, you should use git commit --author $mail-of-author. If pulling
>       changes from somewhere like Poky or OpenMoko there is no problem with that
>       but mention where the changes came from.
>     - Include a Signed-off-by: line indicating the change has valid certificate 
>       of origin as per the Linux kernel
> 
> ... which is nowhere near as prescriptive as the rule that you seem to
> be implementing in your script.  Also, the wiki page has an explicit
> statement that "[these] rules are not hard fast rules", which would
> suggest that even such guidance as it does provide is not intended to be
> taken too dogmatically. 
> 
> Personally I think that the current situation of the TSC applying
> pressure to those people who seem to be chronically unable to write
> suitable checkin comments is a perfectly fine way of solving the
> problem.

Probably it could help developers to check the commit messages on their own
by making this as post commit or pre push hook of some sort in their clones .git/hooks

> 
> On a technical level, I think you could probably do without the sed
> subprocess, and "summary = rev.split(1)" doesn't look quite right
> either.

true. I could have got it if I have tested it. The check for : should also
be inverted.