Re: [PATCHv8 5/9] submodule-config: introduce parse_generic_submodule_config

From: Stefan Beller <sbeller@google.com>
Date: Feb 6, 2016, 01:36 UTC
Message-ID: <CAGZ79kY4=wQvK1nn0Nv+qbR+n1=aqtY31EGwoHszG=jHP-iNTg@mail.gmail.com>
In-Reply-To: <20160206012306.GD28749@google.com>

On Fri, Feb 5, 2016 at 5:23 PM, Jonathan Nieder <jrnieder@gmail.com> wrote:

Show 12 quoted lines

> Hi,
>
> Stefan Beller wrote:
>
>> This rewrites parse_config to distinguish between configs specific to
>> one submodule and configs which apply generically to all submodules.
>> We do not have generic submodule configs yet, but the next patch will
>> introduce "submodule.fetchJobs".
>
> Does this mean that options like submodule.fetchJobs can be specified
> in either .gitmodules or .git/config, like the submodule.<name>.*
> options?

oops, yes that would work.

I was diving down the stack when looking where to put the code for submodule.fetchJobs, and as submodule.c::submodule_config missguided me to submodule-config.c as it told me 'anything starting with "submodule." will be parsed in the submodule-config file', so at the time of writing that felt natural to put it there as well.

So generalizing from there, would we expect to have any submodule options with no section in .gitmodules?

>
> I would expect a good value for fetchJobs to be machine-specific.
> Would it work to put the submodule.fetchJobs handling in
> submodule.c::submodule_config alongside fetch.recurseModules instead?

Sure that's possible. Although I would not mind having a preset default from a project. Would it be possible for a project to be malicious with that?

(Answer yes it can, a superproject including itself multiple times and having parallel set to >=2 behaves similar like a fork bomb when the parallelism is not overwritten in the actual .git/config, I would think)

>
> Thanks,
> Jonathan

Thanks, Stefan