jeantil commented on pull request #831: URL: https://github.com/apache/james-project/pull/831#issuecomment-1007568328
I suggest not starting a third discussion on the same topic. as I [wrote in the first discussion](https://github.com/apache/james-project/pull/808#discussion_r780170243) the behavior for this case was not specified in the contracts so we implemented the simplest that passed the contracts :) With that said, implementing the requested behavior is not so difficut. I propose two options depending on the degree of accuracy we require : - attach a monotonously increasing identifier such as a timestamp (or an ObjectId for increased precision) to all enqueued messages and use that in the filter. This is sensitive to clock drift which could lead to some missed messages (not sure if it's an issue) - query the scheduled topic offset on filter creation and attach that to the filter in addition to the out topic offset. filter scheduled messages as they are consumed and before they are requeued in the out topic. a filter would then expire when both the out topic and the scheduled topic's offset have been exceeded. (a simple way to implement this with the current setup is to duplicate the filter actor and the filter command queue and reuse the existing logic for the scheduled queue) -- 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]
