[patch] make the logging header customisable

[Matthieu Patou]

On 05/22/2013 09:52 PM, Matthieu Patou wrote:
> On 05/22/2013 08:53 PM, Andrew Bartlett wrote:
>> On Wed, 2013-05-22 at 09:59 -0700, Matthieu Patou wrote:
>>> On 05/19/2013 10:06 AM, Matthieu Patou wrote:
>>>> Hello,
>>>>
>>>> Find attached an updated version of this, feedback and review welcome.
>>>>
>>>>
>>>> On 06/12/2012 01:27 AM, Matthieu Patou wrote:
>>>>> Hello Volker & all,
>>>>>
>>>>> A couple of months ago we briefly discussed on the possibility to
>>>>> have a header printed even if we log on the stdout because with more
>>>>> than 1 process logging it can be a bit tedious sometime to understand
>>>>> who print what.
>>>>> My understanding was that you were more or less ok with this idea but
>>>>> that instead of adding a simple boolean to say "add headers also when
>>>>> logging to stdout" you proposed me (at least that was my
>>>>> understanding) to be able to specify the header that we want to have.
>>>>> I've been working on this lately and come with the following patch
>>>>> for the 3.5.x branch, it seems to work ok as if the "debug header
>>>>> template" parameter is not specified the header is as it was (which
>>>>> means no header in case of logging to stdout) but if you specify it
>>>>> then it will be as specified.
>>>>>
>>>>> For the moment the patch is for 3.5.14 as it's what we use at my work
>>>>> and it helps us during the QA and dev debug, it didn't apply yet to
>>>>> master but I'll do the job of front port if the idea of the patch is
>>>>> ok (and if it's also at least mostly ok).
>>>>>
>>>>> Matthieu.
>>> Any body on this patch ? should I push it directly to autobuild ?
>> G'Day Matthieu,
>>
>> Sorry for not getting back to you on this one.  The main thought I had
>> about this, at the time and repeated now is:  We hit DEBUG a lot, do we
>> need the extra complexity?
> I'm not sure it's that much more complicated, or I can make it just a 
> bit more complicated when you select the debug template header option.
Comments ?
>> What is the connection between logging with a header to stdout (which
>> was roundly welcomed, except that it caused some odd failure and Jeremy
>> never got the patch in) and this more complex proposal?
> Well I thought of using the standard header but then I realized that 
> maybe you want to be able to configure what we print before each debug 
> line because all the information might not be relevant to you.
>
Should I push to master ?