Thanks for all the feedback. I have opened SLING-4739 and committed
the code to svn. The documentation is also moved to Sling site [2]

> For me a Sling- prefix
> is enough to make it clear that it's a private, Sling-specific,
> header, but I don't have a strong opinion about it.

@Robert - Code now uses Sling-Tracers and Sling-Tracer-Config has header name

> Is there a fundamental reason the LogTracer class must put ServletFilters
> as inner classes  ? It has 8 inner classes which generally makes
> understanding the code harder and leads to the temptation to reference
> internal implementation details across class boundaries. I understand other
> languages have that flexibility by mistake or intention, but the pattern
> when used in Java tends to lead to problems downstream.

@Ian - I prefer use of inner class if the class is not to be exposed
or not to be known to outer world. However I agree in this case it
went bit overboard, so committed code is refactored to simplify the
logic. Hopefully this is simpler!

Chetan Mehrotra
[1] https://issues.apache.org/jira/browse/SLING-4739
[2] http://sling.apache.org/documentation/bundles/log-tracers.html


On Thu, May 21, 2015 at 2:11 PM, Robert Munteanu <[email protected]> wrote:
> On Thu, May 21, 2015 at 11:38 AM, Chetan Mehrotra
> <[email protected]> wrote:
>> On Thu, May 21, 2015 at 2:00 PM, Robert Munteanu <[email protected]> wrote:
>>> AEM -> Sling is self-explanatory,  and the X- prefix is deprecated with
>>> RFC6648 [1]
>>
>> Well not sure. That recommendation is meant for headers which are
>> meant to be used in a wider context. Header which are very much
>> specific to application are better prefixed to indicate that they are
>> custom ones IMHO [2]. So I would prefer naming them X-Sling-Tracers.
>
> Right, it doesn't apply to 'private' headers. For me a Sling- prefix
> is enough to make it clear that it's a private, Sling-specific,
> header, but I don't have a strong opinion about it.
>
> Robert

Reply via email to