Good morning,
I am writing a language parser, and am looking for the best structure to use for a rollback cache which currently does the following:
- When requesting a new character from the stream, the character is added to the cache, in case a rollback is requested.
- When a rollback is requested, go back to a certain point in the cache so that when another character is requested, it gets it from there instead.
- When a token is found, remove everything in the rollback cache up to the current position.
So in short, I would love to know which you feel to be the best data structure for:
- Priority 1: appending characters (codePoints would be a welcome addition)
- Priority 2: Doing a substring (such as StringBuilder.delete(...)) on the data structure (or clearing completely)
- Priority 3: Being able to create a string of the cache (e.g. StringBuilder.toString())
I hope to hear from you soon!
If I were you, for such a specialised use and with possible performance and resource constraints, I'd implement my own buffer from primitives. I think it's more trouble adapting existing structures. Of course, if it didn't hurt, I'd try to conform to the well known relevant interfaces, such as CharSequence
, Appendable
, List
, etc.