[U-Boot] Problem with first use of patman [FIXED]

[Albert ARIBAUD <albert.u.boot@aribaud.net>]

Hi Simon,

On Sun, 14 Oct 2012 16:53:03 -0700, Simon Glass <sjg@chromium.org>
wrote:

> Hi Albert,
> 
> On Sun, Oct 14, 2012 at 11:52 AM, Albert ARIBAUD
> <albert.u.boot@aribaud.net> wrote:
> > Hello Simon and the list,
> >
> > I'm trying to start using patman, and -- probably out of sheer
> > ignorance -- cannot make it work.
> 
> Not a good first experience, sorry!
> 
> >
> > I've defined my .patman file with aliases for mvgbe, lacie_kw, arm,
> > arm926ejs etc.
> >
> > I am trying to generate patches from a local branch (which I have made
> > available on the u-boot-arm repo as 'add-ws-support') which has two
> > commits.
> >
> > Top commit contains some matman tags, but the error is the without any
> > tag.
> >
> > The complete error is:
> >
> > albert at lilith:~/src/u-boot-arm$ tools/patman/patman -n -c2
> > Traceback (most recent call last):
> >   File "tools/patman/patman", line 125, in <module>
> >     series = patchstream.GetMetaData(options.start, options.count)
> >   File "/home/albert/src/u-boot-arm/tools/patman/patchstream.py", line
> > 352, in GetMetaData ps.ProcessLine(line)
> >   File "/home/albert/src/u-boot-arm/tools/patman/patchstream.py", line
> > 190, in ProcessLine self.commit.subject = line
> > AttributeError: 'NoneType' object has no attribute 'subject'
> > albert at lilith:~/src/u-boot-arm$
> >
> > Can Simon (or someone who's already bumped into this error) tell me what
> > I'm doing wrong?
> 
> I am really not sure about this. I can't seem to repeat the problem
> with your branch.
> 
> I am pretty sure from your backtrace that it is parsing the 'git log'
> output, and seeing a subject before it has seen a commit. It is
> possible that your 'git log' output is non-standard? Perhaps you could
> post that to the list?
> 
> It expects to see:
> 
> commit <hash>
> Author: ...
> Date: ...
> <blank line>
>     commit subject
> <commit lines if any>
> commit <hash>
> Author ...
> 
> 
> You could put a 'print line, self.state' at the top of ProcessLine()
> in patchstream.py and that might help me debug it.

Thanks! Your hint about 'git log' made me find the cause of the problem.

My git log content is in no way special... except it has color ( 'ui =
always' in my .gitconfig) ! This is what troubled patman: trying with
the color spec in .git config commented out, it works ok.

I have located two places in patman where it calls git log, added
option --no-color to them, verified that it worked with color.ui=auto
in my .gitconfig... then used patman to send these very fixes to the
list. :)

> Regards,
> SImon

Amicalement,
-- 
Albert.