those copy and paste bugs are a bit embarrassing :/
both methods have currently no unit test. I will fix that tomorrow.
btw since you helped me spotting bugs already a few times, would you be
interested in doing some api review of more experimental areas?
On 09/28/2011 12:20 AM, notzed [via jogamp] wrote:
> Hi again Michael,
> Spotted another typo, didn't seem worth a real bug report:
> int ret = cl.clEnqueueCopyImageToBuffer(ID, dstBuffer.ID, srcImage.ID,
> ibA, ibB, dstOffset,
> conditions, conditionIDs,
> events==null ? null : events.IDs);
> dstBuffer.ID and srcImage.ID should be swapped.
> On nvidia/gpu I don't know what it did (didn't copy anything i presume), but
> on amd/cpu it spat with an invalid object error.
> FWIW github's source code browser now runs like a pig and makes my laptop
> hot ...
Yeah, they're a bit obvious :) At least when i hit a problem I know where to look ... After a over a year it seems odd that such a function was never hit by anyone, but then again it took me that long to hit it, and even then it was just a work-around for some code.
I'm pretty much sticking to the basic API and my own utilities - and of a not-too-up-to-date build for that matter, but I can have a look if you like. I am using it actively at the moment although very much focussed on the opencl code.