On 08/29/2017 10:08 AM, oliver wrote:
hi, two questions about this:
1.) how does this memory leak show in a patch that has a lot of net traffic (i.e. sends many numbers fast to sync video to a timecode) ?
the memory leak is about leaking connections, so it is only relevant if you do *lots* of (re)connects. the actual transmitted data does not leak.
2.) would it be better to use "iemnet" objects instead ?
as the author of iemnet, i'd say "of course". as a Pd-purist, i'd say it's *far* better to to fix the *vanilla* objects.
at least, with Pd-0.48 you can use iemnet together with fudiparse/fudiformat to build a fully compliant wrapper for netsend/netreceive, so you can have a temporary workaround until the core objects are fixed.
btw, the error in netsend has been fixed in the update/0.48-0 branch, but netreceive still leaks some memory on each reconnection...
gamrds IOhannes