On Thu, 2015-08-20 at 10:46 +0300, Michael Ivanov wrote: > Sorry, > > A couple of months ago I discovered a case in proton that prevented > using more than 32767 nodes in message data. One of our applications > created large messages and wherever it exceeded this limit it just > crashed. The data type used to keep number of nodes in proton is > unsigned short and actually allows to have twice as much number of > nodes (65535), but because of the way pni_data_grow function is > implemented the upper half of the complete value range > (32768...65535) > can never be used. > > To allow our application to run I created a fix and currently use > custom proton version. I also published the fix on mailing list and > created a jira issue on it on mr. Alan Conway's advice, but I see > that version 0.10 still uses the same approach. > > Is there any special reason to limit the data size to 32767 nodes? > It is sad that I have to proceed to use patched version for our > systems.
Sorry Michael, that is my fault. You posted the fix for this some time ago and I didn't spot the JIRA till just recently so it didn't make the 0.10 release. It is on master now and will be in the next release, I believe there will be a 0.10.1 shortly. https://github.com/apache/qpid-proton/commit/0ca6ba7e261d82ac490ef0ea41 1d7a6e37de4591 --------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
