jonahaapala opened a new pull request, #3173: URL: https://github.com/apache/thrift/pull/3173
I noticed that when partial deserialization was added a new method `readFieldBeginData()` was added to elide the TField object creation. However, only the **TBinaryProtocol** takes advantage of this. This PR updates **TCompactProtocol** to do the same. Also, while using the **TProtocol** helper method `readMap()` that takes a **ReadMapEntryCallback** I found myself wishing I could passing a separate keyCallback and valueCallback so my calling code can use method references/lambdas. Another minor tweak in this area was the default `mapCreator` for maps and sets. In later versions of Java they added the `HashMap.newHashMap()` factory method which will size the table to prevent resizing when given an initial capacity that won't be exceeded. I don't know of an equivalent for HashSet, though the HashSet constructor that takes in a Collection also performs this optimization. I have added them here for convenience. I have noticed in the past that the Java thrift compiler initially sizes maps to 2x their size likely for this reason, so this is like an analogue to that. Perhaps the compiler should also be doing this, but I suspect it's of minor consequence either way. While trying to utilize the **TProtocolDecorator** I was sad that none of the `skip*()` methods were delegated to. I mostly wanted `skipBinary()`, but added them all for completeness. Again, looks like these were added for partial deserialization so I assume it was just missed. I added a few more overrides to it as well to really treat it as a delegating wrapper. -- 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: dev-unsubscr...@thrift.apache.org For queries about this service, please contact Infrastructure at: us...@infra.apache.org
