ppkarwasz commented on issue #3792: URL: https://github.com/apache/logging-log4j2/issues/3792#issuecomment-3035412560
Hi @fabian-braun, The core issue with `lastEpochInstantAndBuffer` is that it's **not immutable**, unlike the previous `CachedTime` implementation: https://github.com/apache/logging-log4j2/blob/054a4f7084b8a03f932bc9dd40ebba55142a36be/log4j-core/src/main/java/org/apache/logging/log4j/core/pattern/DatePatternConverter.java#L161-L171 Using a `String` instead of a `StringBuilder` would break the GC-free behavior expected when `log4j2.enableThreadlocals == true`, so we need to proceed carefully. Here’s what I propose: 1. **In version 2.25.1** (scheduled for release next week), we remove `lastEpochInstantAndBuffer` entirely and eliminate inter-thread sharing. We’ll rely exclusively on the thread locals. 2. **In version 2.26.0**, we could consider introducing a non-GC-free version of `InstantPatternThreadLocalCachedFormatter`. This variant would be used when `log4j2.enableThreadlocals = false`, and it would format timestamps to immutable `String` values. @vy What do you think? Since this is the last urgent issue for `2.25.1`, I’ll go ahead and open a PR for the first point shortly to help speed up development. That way, we can review the three open PRs over the weekend and aim to stage `2.25.1` next week. -- 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: notifications-unsubscr...@logging.apache.org For queries about this service, please contact Infrastructure at: us...@infra.apache.org
