Re: Allow pidfile(3) to create pidfiles in arbitrary locations

[Julio Merino <jmmv%julipedia.org@localhost>]

On Tue, Mar 29, 2011 at 1:44 PM, Julio Merino <jmmv%julipedia.org@localhost> 
wrote:
> On Tue, Mar 29, 2011 at 1:46 PM, David Laight <david%l8s.co.uk@localhost> 
> wrote:
>> On Tue, Mar 29, 2011 at 09:26:01AM +0100, Julio Merino wrote:
>>>
>>> On a second thought, I have updated the change to consider paths with an
>>> explicit '.pid' extension to be considered relative.  Otherwise, I'd expect
>>> confusion if the value to pidfile(3) was provided blindly from a
>>> user-provided parameter (like in foo -p foo.pid) and the pid file was 
>>> created
>>> in /var/run.  The "only" way to get the old behavior is to just provide a
>>> basename (e.g. foo -p foo).
>>
>> Probably be better to assume /var/run unless the filename contains a '/'.
>> So you'd need -p ./foo.pid to get a file in the current directory.
>
> That's what the original patch did, but it's quite confusing.  (I'll
> also say that the .pid heuristics I just added may be even worse...)

You know what... I'll just revert these early-morning, not-awake-yet
changes and submit the patch in its previous form.

-- 
Julio Merino / @jmmv