I've gone through and changed interface_id to a guint32, along with
max_interface_id, leaving interface_count (I didn't see the need for it to
be increased).
I've forked the repo and have it over at
https://github.com/tastywheattasteslikechicken/mono/commit/6e7f1d94eb07822fb7bf5687b67df5076f969818
disregard email.
Thanks,
Greg
On Tue, Jan 5, 2016 at 5:28 PM, Greg Young wrote:
> In dealing with very tight code and measuring mono we have found that
> even registering empty handlers from a profiler can cost about 50% of
> performance. Is there (or are there plans) for allowing a profiler to
In dealing with very tight code and measuring mono we have found that
even registering empty handlers from a profiler can cost about 50% of
performance. Is there (or are there plans) for allowing a profiler to
selectively register to allow for better performance? From
benchmarking here a key one wo
Yeah, what I'm chasing are the best ways to cause failures without this
assertion -- that way as I go through and fix things I can be more
confident that things haven't been missed.
On Tue, Jan 5, 2016 at 11:45 PM, Alexander Köplinger <
alexander.koeplin...@xamarin.com> wrote:
> I don't think it'
I don't think it's as easy as removing the assert.
For one, the interface ID returned by mono_get_unique_iid() is stored as a
guint16 (
https://github.com/mono/mono/blob/36c7332104eb5250a93079ae77c2e0dbf12c6c9a/mono/metadata/class-internals.h#L344)
which means you only have 0-65,535 values there.
