From: Junio C Hamano <gitster@pobox.com>
Date: Wed, 10 Feb 2016 02:04:45 GMT
Subject: Re: [PATCHv9 3/6] fetching submodules: respect `submodule.fetchJobs` config option
Message-ID: <xmqqtwlhz5fm.fsf@gitster.mtv.corp.google.com>
In-Reply-To: <xmqqy4atz7u0.fsf@gitster.mtv.corp.google.com>

Junio C Hamano <gitster@pobox.com> writes:

> I actually do not think we mind git_parse_int(), git_parse_long(),
> and git_parse_uint() to complement git_parse_ulong().  I am not
> enthused by the "nonnegative-int" thing, though.
>
> Do we have enough cases where we want to use signed type and reserve
> negative value for our own internal use (e.g. "unspecified yet")?
> If not, a very generic git_config_int() with a caller specific range
> check wouldn't look _so_ bad.
>
> 	parallel_jobs = git_config_int(var, val);
>         if (parallel_jobs < 0)
> 	     	some corrective action;
> 	return 0;

And of course, if we _do_ have many callsites where the caller wants
a sub-range of the range allowed by the underlying type, we can have

 	<T> git_config_<T>_bounded(var, val, <T> lb, <T> ub)

where T are the usual integral types and lb and ub are (inclusive)
lower and upper bounds.  With that, Jonathan's non-negative-int
would fall out as a natural consequence:

	parallel_jobs = git_config_int_bounded(var, val, 0, INT_MAX);

I vaguely recall that Michael Haggerty had a series to add and use
helpers to parse values of integral types and something like that
interface was suggested in the discussion; and it may even have used
a word better than "bounded".