My guess would be the site manual. 
So we should be careful in how much detail to reveal on the site manual. 

For example, I deliberately did not write the path and file name of the plugins 
metadata file (Log4j2Plugins.dat). 

I noticed that the full path of that file is mentioned in the plugins manual 
page after applying Scott's LOG4J2-745 patch. Would it not be better to keep 
that an undocumented implementation detail ?

Sent from my iPhone

> On 2014/08/07, at 13:46, Matt Sicker <[email protected]> wrote:
> 
> Now when you say manual, do you mean the site manual or the javadocs?
> 
> 
>> On 5 August 2014 08:20, Ralph Goers <[email protected]> wrote:
>> Well, in my view there are two kinds of public. 1) the API users code to add 
>> logging to their code, 2) the API they code to to create Layouts, Appenders, 
>> Filters and other plugins. In short, if itis documented in the manual it 
>> should be considered public.
>> 
>> Ralph
>> 
>>> On Aug 5, 2014, at 6:01 AM, Gary Gregory <[email protected]> wrote:
>>> 
>>> I think the separation we have so far is that log4j-api is "public" and 
>>> everything else not. We can provide more fine grained definitions of course.
>>> 
>>> Gary
>>> 
>>> 
>>>> On Tue, Aug 5, 2014 at 8:55 AM, Matt Sicker <[email protected]> wrote:
>>>> Oh, I know what you mean. However, sometimes documentation is useful for 
>>>> internal use. Is it enough to mark particular classes as internal? Like 
>>>> with an annotation or just in the javadoc or something. Annotation or 
>>>> javadoc tag could be used during site generation to handle them as 
>>>> something special.
>>>> 
>>>> 
>>>>> On 5 August 2014 06:31, Gary Gregory <[email protected]> wrote:
>>>>> Cool. I am disapointed at the lack of docs in some FOSS projects, so it's 
>>>>> good to see someone stepping up. 
>>>>> 
>>>>> That said, keep in mind that some people consider Javadoc to be a 
>>>>> contract definition, and that modifying something that is doc'ed in a 
>>>>> future release is a break in compatibility. 
>>>>> 
>>>>> So let's be careful in what we document and how. I do not think we have 
>>>>> documented what our policy is on that though. I'm am just worried that if 
>>>>> we doc the internal workings of some method or class, users will then 
>>>>> rely on that behavior and complain if it changes. Aside from that, doc 
>>>>> away! 
>>>>> 
>>>>> Gary
>>>>> 
>>>>> 
>>>>> -------- Original message --------
>>>>> From: Matt Sicker
>>>>> Date:08/04/2014 22:36 (GMT-05:00)
>>>>> To: Log4J Developers List
>>>>> Subject: In regards to javadoc updates.
>>>>> 
>>>>> So as you may have already noticed, I like to write a lot. This carries 
>>>>> over to documentation, and I prefer javadocs when it comes to that as 
>>>>> it's far easier to keep in sync with how the code works. With that in 
>>>>> mind, I've been elaborating about random classes to help improve the 
>>>>> docs. However, since I'm obviously not the author of everything here by 
>>>>> any means, I might misinterpret classes or functions slightly. If I do, 
>>>>> please let me know!
>>>>> 
>>>>> As a developer, I usually skip straight to the javadocs for any library I 
>>>>> use, and I'm oftentimes disappointed by the lack of documentation in 
>>>>> them. Examples can become stale over time. A good majority of the JDK is 
>>>>> well documented (there are several exceptions of course), and I like to 
>>>>> try to follow that pattern myself.
>>>>> 
>>>>> -- 
>>>>> Matt Sicker <[email protected]>
>>>> 
>>>> 
>>>> 
>>>> -- 
>>>> Matt Sicker <[email protected]>
>>> 
>>> 
>>> 
>>> -- 
>>> E-Mail: [email protected] | [email protected] 
>>> Java Persistence with Hibernate, Second Edition
>>> JUnit in Action, Second Edition
>>> Spring Batch in Action
>>> Blog: http://garygregory.wordpress.com 
>>> Home: http://garygregory.com/
>>> Tweet! http://twitter.com/GaryGregory
> 
> 
> 
> -- 
> Matt Sicker <[email protected]>

Reply via email to