Any further thoughts on this from the other devs given the recent mail on
logging?

Colm.

On Mon, Apr 11, 2016 at 6:59 PM, [email protected] <[email protected]> wrote:

> @Colm, this is an important. I appreciate you highlighting it.
>
> IMO We should standardize on slf4j as most hadoop projects seem to use.
> In most environments that we run in slf4j is available.
> - hfs
> - hbase
> - hive
> - storm
> - knox
> - kafka
> - solr
>
> For webapps
> - kms explicitly includes it.
> - While admin does not include it explicitly it must be getting it because
> of transitive dependencies as a few files in admin are are currently using
> slf4j.
>
>
> Ideally the following should happen:
>
> - Standardize on one mechanism, i.e. replace all uses in favor of the one
> standard mechanism.
> - find a mechanism to flag alternate means of doing logging and fail the
> build, perhaps via one of the static analysis tools so this does not happen
> in the future.
>
> What do others feel?
>
>
> - Alok
>
>
>
> On 3/30/16, 1:55 AM, "Colm O hEigeartaigh" <[email protected]> wrote:
>
> >Hi all,
> >
> >Ranger uses many different logging mechanisms:
> >
> >find . -name "*.java" | xargs grep -l "slf4j" | wc -l
> >30
> >find . -name "*.java" | xargs grep -l "log4j" | wc -l
> >125
> >find . -name "*.java" | xargs grep -l "commons.logging" | wc -l
> >166
> >
> >IMO we should standardise on a single method for consistency. Any ideas on
> >which? I would tend to favour SLF4J personally.
> >
> >Colm.
> >
> >
> >--
> >Colm O hEigeartaigh
> >
> >Talend Community Coder
> >http://coders.talend.com
>



-- 
Colm O hEigeartaigh

Talend Community Coder
http://coders.talend.com

Reply via email to