Aleksey Bragin wrote:
Also, as in discussion in bug #4169 I see total lack of understanding. Johannes fixes how he thinks it's right, Dmitry fixes another place thinking he is right, Stefan blames everyone thinking he is right.
What anyone of you have done to ensure correct behaviour? Was testing of the kernel exported routine has been done? And then testing advapi's code above the "correct" implementation? Even if the current kernel's implementation is bad, it's not a reason to hack it further.
Also asking in the ros-dev ML helps if any of you have no idea what they are fixing. There are still some knowledgable people here.
This regression, which is known for 4 months and about which people complained on the forum, I thought it was enough reason to return to a hack when nobody is working on it.