gaborgsomogyi edited a comment on pull request #29964: URL: https://github.com/apache/spark/pull/29964#issuecomment-708226412
@HyukjinKwon The main intention is to make the framework as self-serving as possible. I see mainly 2 use-cases (I'm sure others can came up more): * The built-in provider works fine for a specific database (for example MSSQL) but the user wants different authentication functionality which is not able to be done with the built-in one. Such case the user implements a new provider but when added then 2 providers will be in place to handle MSSQL authentication (the binding is class name based). Such case the built-in one must be turned off to use the new one. * The built-in provider contains a bug for a specific database. Such case full Spark patch build can be provided in normal case but this configuration opens new ways in testing and patching. Namely one can easily send a custom provider to proof/solve provider related issues (of course temporarily). As a general opinion all the issue solutions and good features must be added upstream but it could take some time so having such flexibility would pay off in terms of reaction time. ---------------------------------------------------------------- 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. For queries about this service, please contact Infrastructure at: [email protected] --------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
