Whitespace at the start of first line of commit

[Joel Brobecker]

> A quick run of the testsuite reveals that this assumption is made
> all over. I am opposed to having this feature be a standard feature
> of the git-hooks, so you wouldn't have to add an ad hoc check.
> The way I would do it is by enhancing the git_run function to check
> for a parameter named "_no_strip" and block the strip() operation
> when passed as True. You can then implement your check using the usual
> git interface, with the extra "_no_strip=True" parameter.
> 
> Do open a GitHub issue if you'd like me to add this check to
> the git-hooks. I will likely give it less priority because
> you'll have a way to implement it on your on, but I will get to it
> eventually.

I apologize.  Re-reading myself, I'm mixing various issues and not being
clear as a result, let me rephrase:

   For your immediate need, I am proposing a way to adapt the git_run
   function so you can use it to implement your check. Another option
   is to use Python's subprocess module directly, but I think this will
   be more work than what I suggest.

   For the longer term, you'll have access to a hook that you can
   use to call a script to validate an update. In that hook, you'll
   have total implementation freedom, so you can implement that check
   there, and not have to worry about that implementation detail in
   the git-hooks.

   Also longer term, I can add that check directly in the git-hooks
   as well. I don't remember anyone ever really making this kind
   of mistake, but it doesn't mean that it's unlikely either, nor
   that it would be specific to the GCC community. Hence the merit
   of having it in git-hooks directly.

-- 
Joel