Matěj Laitl

Hi all, sorry for replying so late - too little time lately, but I've been keeping an eye on this. Has anyone tried with the current git version? It has...

Just to share info: this should be solved once @Dapid's pull request is merged.

Thanks for this pul request, I'll look at the commits more closely as soon as I find some time. One thing to note now, however: would you mind changing the...

Thanks, @Dapid - sorry for being extremely slow to respond, I've been rather busy lately, you know that.. I'll have a look at the latest version when time permits.

> I get the same on Linux. It is because the timing module in the benchmark is not thread safe, and cannot be used inside prange. I am working on...

Hi, this seems to be caused by the fact that setup.py is still written for Python 2.7, you seem to be installing under Python 3.x. Should be easy to fix...

Hmm, seems to be a mismatch somewhere between a vector (1D array) and 1-row matrix (2D array). Can you print `shape` of `yt` at `RFID/pr_model/scripts/frequencyModel.py", line 151, in addObservation`?

Hi, this breaks build with Cython and also totally breaks PyBayes with Python 2.x and Nupy different than 1.9. This must be solved differently.

> Unfortunately for some reason, Apache 5 is not starting in the test runs - I think this is to do with matching of the canonical hostname in the creation...

For reference: there is a [work-around to this mentioned in serde_qs docs](https://docs.rs/serde_qs/0/serde_qs/index.html#flatten-workaround), which should apply equally to serde_urlencoded. The work-around has been mentioned in the equivalent issue in serde_qs: https://github.com/samscott89/serde_qs/issues/14#issuecomment-456865916