Re: [PATCH v3 2/3] Introduce GIT_MMAP_LIMIT to allow testing expected mmap size

From: Steffen Prohaska <prohaska@zib.de>
Date: Aug 24, 2014, 16:03 UTC
Message-ID: <E4460202-E37A-498E-8062-DCCD9AE8E38C@zib.de>
In-Reply-To: <xmqqiolkpjtq.fsf@gitster.dls.corp.google.com>

On Aug 22, 2014, at 6:31 PM, Junio C Hamano <gitster@pobox.com> wrote:

Show 15 quoted lines

> Steffen Prohaska <prohaska@zib.de> writes:
> 
>>>> +	if (limit == -1) {
>>>> +		const char *env = getenv("GIT_MMAP_LIMIT");
>>>> +		limit = env ? atoi(env) * 1024 : 0;
>> 
>> ... this should then be changed to atol(env), and ... 
> 
> In the real codepath (not debugging aid like this) we try to avoid
> atoi/atol so that we can catch errors like feeding "123Q" and
> parsing it as 123.
> 
> But it is OK to be loose in an debugging aid.  If I were doing
> this code, I actually would call git_parse_ulong() and not
> define it in terms of kilobytes, though.

Thanks for suggesting this. I wasn't aware of git_parse_ulong(). I think it makes very much sense to use it, even though it's only a testing aid.

I'll send a PATCH v5 series.

	Steffen