[slf4j-dev] svn commit: r1086 - in slf4j/trunk/slf4j-api/src: main/java/org/slf4j/helpers test/java/org/slf4j/helpers

[Jörn Huxhorn]

On Fri, Aug 1, 2008 at 9:29 PM, Ceki Gulcu <listid at qos.ch> wrote:

>
>
> The probability of passing a cyclical array is very low. Moreover, if and
> when
> it happens, the user will immediately know the cause of the failure. The
> StackOverflowError will point to MessageFormatter.deeplyAppendParameter
> method,
> so it will be easy to identify. If the bug were hard to identify, then it
> would
> have been an entirely different matter. At this stage, I am just too lazy
> to be
> bothered.
>

Are you serious?

I once heard a lecture on logging and the prof said that the main objective
of any logging framework must be to not break the application it is logging.

>From this point of view - that I absolutely share - the quality of the slf4j
framework would be lower after applying toString for arrays than before
since now there is a real possibility to crash an application by simply
issuing a log message.

It's also not our task to define that recursive arrays are a bug. With the
previous implementation a log message containing such an array would have
worked, with the new implementation it would crash the application.

Whether something like recursive arrays makes sense is purely application
specific.

Beside that, the fix is really simple...
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://qos.ch/pipermail/slf4j-dev/attachments/20080803/86a681d8/attachment.htm>