As far as I understood, such changes affect mainly configuration (timeouts, limits, offsets, etc), right? Would not be better to submit a PR to the upstream project to allow to customize such configurations without modifying the code?
On Tue, Sep 27, 2016 at 8:49 AM, Jong Wook Kim <[email protected]> wrote: > Related to the recent comments raised in the vote thread, I’d like to > revisit the asynchbase issue. > > Now their fixes on NSRE are tagged in the recently fixed 1.7.2 on Maven > Central, the remaining differences between our custom version and the > official version are: > > - RPC-wise timeout setting in Scanner > - limit and offset setting in GetRequest > > I made a small patch, as seen in here <https://github.com/jongwook/ > incubator-s2graph/commit/ad5c7f89e46ddbd5dfd9b8721737aa22f94b4002>, which > includes GetRequest.java and Scanner.java in the s2core tree along with a > utility that forces loading the bytecode from s2core’s classpath. > > Having two ~1000-line java files which are duplicates might be a bad > practice, but it eliminates the need to maintain a separate codebase and > maven repository for those small patches. > > To be more aesthetically satisfying, I’ve fiddled a little bit with > bytebuddy <http://bytebuddy.net/#/> to make runtime modification of the > behaviors of those classes, but the modifications are kindof scattered > making it harder to write the dynamic proxy. > > I’d appreciate any comments on the patch above. > > JW -- Sergio Fernández Partner Technology Manager Redlink GmbH m: +43 6602747925 e: [email protected] w: http://redlink.co
