Makes sense, I will leave it for now. -Jay
On Fri, Sep 14, 2012 at 10:14 AM, Jun Rao <jun...@gmail.com> wrote: > The current semantics of those exceptions are: > InvalidMessageException: message is corrupted (crc doesn't match payload) > InvalidMessageSizeException: message size is larger than the payload size. > This is mostly used on the consumer side in case we fetch a chunk of bytes > on a wrong starting offset. > MessageSizeTooLargeException: If a producer sends a valid message whose > payload is larger than a configured size. > > So, they really have different meanings (yes, the difference is small and > subtle). > > Thanks, > > Jun > > > On Fri, Sep 14, 2012 at 9:55 AM, Jay Kreps <jay.kr...@gmail.com> wrote: > > > This seems a bit duplicative. What if we removed > > MessageSizeTooLargeException and used the existing > > InvalidMessageSizeException instead--no need for both. I would also like > to > > make InvalidMessageSizeException subclass InvalidMessageException. > > Thoughts? > > > > -Jay > > >
