Austin Wright

Wouldn't a second `xmlns:p2` (on the p2:fourth element) be redundant, here?

Oh for some reason I was reading that completely backwards; yeah, the specified XMLDOM output isn't correct.

Yeah I have this problem too, though usually only when I haven't properly handled an error somewhere else. I haven't been able to reliably reproduce it, it just sort of...

I've only been able to reproduce it after another error. I'm suspecting there may be a missing `var` or similar, somewhere. But I can't see anything obviously wrong with the...

If only half a message of the stream were to come in (which is perfectly legal), what would happen here? https://github.com/marcello3d/node-mongolian/blob/f1a6bdc547db1021321acc869d530ac05dda24d3/lib/connection.js#L59-L79 It appears that there's no test to make sure...

So this probably is the problem then? I'm currently re-writing the file to avoid Buffer re-allocations, among other things

That might be the case for the second and beyond errors ("Expected 1701080942") but the first one seems to simply be an incomplete packet, with no prior error having been...

Yeah, that's exactly the problem: whenever an Exception is thrown in an emitter hook, it interrupts the stream handling and an entire segment goes unparsed. So, how should errors be...

Should it really be disconnecting on error by default: https://github.com/marcello3d/node-mongolian/blob/f1a6bdc547db1021321acc869d530ac05dda24d3/lib/server.js#L212-L215 ? Here's my file that seems to fix: https://gist.github.com/Acubed/5167758 And available in commit Acubed/node-mongolian@2b48848c31cc80c4d0b18483cd401eeea1b9033d

Perhaps the event should be called with process.nextTick? That would isolate any errors from affecting the main loop. I'm not so sure if it should disconnect from the database, but...