Re: [rfc] autoconf-refactor: change semantics of valgrind option?

On Mon, Jun 7, 2010 at 7:12 AM, Amos Jeffries <squid3_at_treenet.co.nz> wrote:
> Kinkie wrote:
>>
>> Hi all,
>>  we currently have a non-standard way of handling valgrind: instead
>> of the common --enable-valgrind, we now use a --with-valgrind[=path].
>> I'd like to change that to the standard behavior of a boolean --enable
>> flag, the path can be passed using environment variables (CPPFLAGS).
>>
>> This changes the user-visible behavior slightly, I'd like to gather
>> feedback.
>>
>> Thanks
>>
>
> The "standard" we try to use is the GNU autoconf recommendations.
> They outline --with-PACKAGE when support or linkage of some other external
> software is to be included, and --enable-FEATURE for specific local
> features.
>
> IMO valgrind is an external package we optionally link in special support
> and definitions for.

Well, what I recall from the discussion is that valgrind support for
squid acts by changing the memory allocation mechanisms to be
valgrind-friendly. But I'm okay with any decision :)

-- 
    /kinkie