Re: [PATCH 3/4] git-commit-graph.txt: refer to "*commit* graph file"

["Martin Ågren" <martin.agren@gmail.com>]

On Thu, 20 Sep 2018 at 14:50, Derrick Stolee <stolee@gmail.com> wrote:
>
> On 9/19/2018 12:30 PM, Martin Ågren wrote:
> > The full name, by the way, is not the "commit-graph file" with a dash,
> > cf. the synopsis. Use the dashless form. (The next commit will fix the
> > remaining few instances of the "commit-graph file" in this document.)
>
> The file is literally at ".git/objects/info/commit-graph" which is why I
> tried to use "commit-graph" everywhere. Why do you think that "commit
> graph" is better?

I noticed the discrepancy between "commit graph file" and "commit-graph
file" and briefly wondered if it was intentional, i.e., if it meant
anything, but the dash vs no dash seemed pretty random to me. In order
to figure out which was (more) correct, I went to the synopsis. But
admittedly, that was quite arbitrary. For all I know, "the commit-graph
file" could be the better choice, grammatically.

There is the file named "commit-graph" as you note, but it might on the
other hand just as well be called "cg.bin". I would probably try to let
the filename "commit-graph" influence the user manual only if we would
have written "cg.bin" instead. For example, if we would talk about how
you might get out of a hole by deleting the "<...>/commit-graph"
("cg.bin") file manually.

But that's certainly not to argue against "the commit-graph file". I'd
be happy to s/commit graph file/commit-graph file/g instead to keep
others from wondering if these are two slightly different things. And
if the concept and the file have the same name, all the better.

If you agree, I'll do that in a v2, where I will also note in the
Options section that `--object-dir` takes a `<dir>`.

Martin