Closed tnowotny closed 5 years ago
Oh, I'm sorry about that, this is due to the changes in brian-team/brian2#1111. I actually already have a fix, but it was written on top of the changes in #92 , so I wanted to wait until that one got merged. Maybe we should just do that now? I was waiting for a confirmation that the fix is ok, but everything worked correctly for me at least.
Yes, I am not sure we can wait for @schlowm0 to approve the pull request ... so probably ok to merge it now and hope it is ok.
Hi @mstimberg ... I am still getting the original test failures as described above ... is there still another issue here?
Ah ... and scrolling through the entire trace of error output, not surprisingly the test that is failing is actually called brian2.tests.test_functions.test_binomial
That PR (#94) isn't merged yet, or do you mean that you still get the errors when you check out the fix_vectorisation_idx
branch?
oh, sorry ... I jumped the gun there. I didn't realise it wasn't merged yet.
Sorry guys I recently switched labs and I'm still in the process of setting up
Hi @mstimberg , I was trying to look into the issues we had with benchmarking and was going through a few things when I noticed that the Brian tests didn't seem to run through. I am a bit clumsy with this stuff, so was not quite sure how I find out which test it is that is failing. But the failure essentially boils down to that code is generated in the neuronFnct or neuronKrnl that calls the support_code functions _binomial_2 and _binomial_3 without arguments whereas those functions are expecting a seed argument of type int64_t. Have you come across this before? I am using manual check-outs for brian2 and brian2genn, both at the latest master.