on 2/19/02 11:47 AM, "John Wilson" <[EMAIL PROTECTED]> wrote:
> Have you done any performance testing on this new code? Historically we had > quite a bit of problems with people reporting very poor performance with > large (multi megabyte) Base64 values (Yes, people do do this sort of thing > with XML-RPC!). I think Hannes did some tuning work on the old code. > > John Wilson > The Wilson Partnership > http://www.wilson.co.uk My decision to use Catalina' code is based entirely on the fact that Catalina is *by far* the most used piece of software on Jakarta. Something like 100-200,000 downloads/week. If the Base64 implementation sucked, it would have been discovered a long time ago. That said, we *have* to get rid of the LGPL one. We just can't have that in CVS. It was a terrible idea to spend time tuning it. It should have been swapped out a long time ago. Note, I did introduce a potential bottle neck in the xmlrpc code in that in one case, I had to allocate a String in order to convert from byte[] to char[]. If you are willing to write an implementation of Base64 that outputs a char[] directly (like the GPL one did), that would be great. At CollabNet, we aren't passing binary data around over XML-RPC, so we won't be hitting that code. -jon
