Dandandan commented on PR #24562: URL: https://github.com/apache/datafusion/pull/24562#issuecomment-5423374538
> I haven't taken a super close look at the code but I like the idea. Im thinking of how quickly this could become cost inefficient > > ``` > Currently it is opt-in during development: when adding the performance label it runs the TPC-H SF=10 benchmark on each commit. > ``` > > with how big DataFusion is we can possibly have several engineers hamming this script and that can become expensive fast. > > Also what if the PR needed flip a flag before running benchmarks. (the PR could also enable this by default before pushing, so this isn't too important) Re expensive: I think the largest cost currently is just building the binaries (benchmarks itself finishes quicker). Doing it on a label + merge only tries to lower the expense. -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. To unsubscribe, e-mail: [email protected] For queries about this service, please contact Infrastructure at: [email protected] --------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
