Web lists-archives.com

Re: [PATCH 2/2] t3430: update to test with custom commentChar




Hi Daniel,

On Mon, 9 Jul 2018, Daniel Harding wrote:

> On Mon, 09 Jul 2018 at 00:02:00 +0300, brian m. carlson wrote:
> > On Sun, Jul 08, 2018 at 09:41:11PM +0300, Daniel Harding wrote:
> > > Signed-off-by: Daniel Harding <dharding@xxxxxxxxxxxxx>
> > 
> > > diff --git a/t/t3430-rebase-merges.sh b/t/t3430-rebase-merges.sh
> > > index 78f7c9958..ff474d033 100755
> > > --- a/t/t3430-rebase-merges.sh
> > > +++ b/t/t3430-rebase-merges.sh
> > > @@ -56,12 +56,12 @@ test_expect_success 'create completely different
> > > structure' '
> > >    cat >script-from-scratch <<-\EOF &&
> > >    label onto
> > >   -	# onebranch
> > > +	 > onebranch
> > >    pick G
> > >    pick D
> > >    label onebranch
> > >   -	# second
> > > +	 > second
> > >    reset onto
> > >    pick B
> > >    label second
> > 
> > Should this affect the "# Merge the topic branch" line (and the "# C",
> > "# E", and "# H" lines in the next test) that appears below this?  It
> > would seem those would qualify as comments as well.
> 
> I intentionally did not change that behavior for two reasons:
> 
> a) from a Git perspective, comment characters are only effectual for comments
> if they are the first character in a line
> 
> and
> 
> b) there are places where a '#' character from the todo list is actually
> parsed and used e.g. [0] and [1].  I have not yet gotten to the point of
> grokking what is going on there, so I didn't want to risk breaking something I
> didn't understand.  Perhaps Johannes could shed some light on whether the
> cases you mentioned should be changed to use the configured commentChar or
> not.
> 
> [0]
> https://github.com/git/git/blob/53f9a3e157dbbc901a02ac2c73346d375e24978c/sequencer.c#L2869
> [1]
> https://github.com/git/git/blob/53f9a3e157dbbc901a02ac2c73346d375e24978c/sequencer.c#L3797

These are related. The first one tries to support

	merge -C cafecafe second-branch third-branch # Octopus 2nd/3rd branch

i.e. use '#' to separate between the commit(s) to merge and the oneline
(the latter for the reader's pleasure, just like the onelines in the `pick
<hash> <oneline>` lines.

The second ensures that there is no valid label `#`.

I have not really thought about the ramifications of changing this to
comment_line_char, but I guess it *could* work if both locations were
changed.

Ciao,
Johannes