On Tue, Jun 20, 2017 at 8:45 PM, Jonathan Tan <jonathantanmy@xxxxxxxxxx> wrote:
> Thanks, this looks like a good explanation. Some more nits, but
> overall I feel like I understand this and have learned something from
> it.
>
> On Tue, Jun 20, 2017 at 3:56 PM, Stefan Beller <sbeller@xxxxxxxxxx> wrote:
>> +A submodule is another Git repository tracked inside a repository.
>> +The tracked repository has its own history, which does not
>> +interfere with the history of the current repository.
>> +
>> +It consists of a tracking subdirectory in the working directory,
>> +a 'gitlink' in the working tree and an entry in the `.gitmodules`
>
> Probably should be `gitlink` (the special quotes), and (optional)
> s/`gitlink`/`gitlink` object/ because it might not be apparent that
> gitlink is a type of object.
>
>> +file (see linkgit:gitmodules[5]) at the root of the source tree.
>
> After reading below, maybe we should mention the Git directory in
> $GIT_DIR/modules/<submodule name> as part of what a submodule consists
> of too.
I implemented the rest of the suggestions, but this one leaves my head
scratching.
It's not as if we don't know what a submodule is already. Our glossary says:
submodule
A repository that holds the history of a separate project inside
another repository (the latter of which is called superproject).
So by that definition the gitlink and the entry in .gitmodules are just
metadata by the superproject and the actual submodule is the
repository consisting of its working tree (inside the superprojects
working tree)
as well as its git directory preferably inside the superprojects git dir.
This definition however is broken IMHO, as when there is proper meta data
(gitlink + .gitmodules entry), you can have an "uninitialized" submodule,
which has neither a working tree nor a git dir at the time.
The wording above holds true for un-{initialized,populated} submodules
as well, as the tracking directory is empty and the git dir doesn't exist.
So I think I'll resend without that change.
Thanks,
Stefan
