hbase-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Ted Yu <yuzhih...@gmail.com>
Subject Re: Checkstyle News
Date Wed, 02 Dec 2015 23:04:27 GMT
Nice work.

In dev-support/test-patch.sh :
mv target/site/checkstyle.css $PATCH_DIR

This seems to give the following in console:

    Determining number of patched Checkstyle errors.

mv: cannot stat 'target/site/checkstyle.css': No such file or directory
hbase-common/src/main/java/org/apache/hadoop/hbase/codec/BaseDecoder.java	LineLengthCheck
0	1

Just a minor thing.

On Wed, Dec 2, 2015 at 2:46 PM, Apekshit Sharma <appy@cloudera.com> wrote:

> Hey,
> Checkstyles have changed recently. And while they are small changes, they
> do affect the codebase in big way as time progresses.
> New checkstyles were added in HBASE-14829. Description of the jira lists
> the new checks that were added. As of result of these new checks, our
> checkstyle error count increased from ~3.8k to ~17k. :-(
> But as we fix these over time and follow better coding guidelines, it'll be
> all for good.
> Change in testing for checkstyle errors:
> Earlier, we were simply comparing number of errors before and after patch.
> However, there was a simple python tool which Nick wrote, which analyzed
> checkstyle reports to return the name of culprit files. It's clearly a
> better way to detect checkstyle errors. So why not use it instead.
> I updated the tool to check for new errors at <file, type of check> level.
> It runs as a part of test-patch.sh now, and dumps culprit file names, type
> of errors and old & new counts in the console output. (It's not as easy to
> dump exact line numbers, so that granularity, maybe in future).
> Hunting for checkstyle errors is still the same, except that there is no
> need to manually run dev-support/checkstyle_report.py.
> 1. Go to console output, look at list of "filename, check, #old errors,
> #new errors".
> 2. To get exact line number, look for that file in
> 'checkstyle-aggregate.html'
> Follow up issues were:
> HBASE-14859
> HBASE-14890
> HBASE-14902
> - Appy

  • Unnamed multipart/alternative (inline, None, 0 bytes)
View raw message