tuxji commented on pull request #674: URL: https://github.com/apache/daffodil/pull/674#issuecomment-967395784
Some questions for clarification: 1. What is the pitfall of just taking a snapshot of variables’ values like other state and restoring the snapshot when resetting back to the PoU? A snapshot copies variables’ values at a certain point in time and restoring the snapshot resets variables back to their snapshotted values. What does a variable’s mutability have to do with that, that is, why would snapshotting variables be more complicated than snapshotting other state? I do understand the overhead of copying all variables’ values in every snapshot and resetting them (likely back to the same values) when restoring snapshots. 2. Adding copy-on-write behavior to variables sounds like an attempt to optimize away some of the overhead on the assumption that variables change values infrequently. Do you have any range of numbers (even ballpark guesses) for how many variables might exist during a parse (from zero to 20?) and how many times variables really change values during a parse (from zero to 1000 times)? My suggestion is to replace the broken code (which makes assumption) with simple code that is obviously correct in behavior while measuring performance before and after. When you are satisfied the code works correctly, then you can look for ways to use copy-on-write, caches, etc., to speed up performance. -- 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]
