Giordon Stark

Probably inspired by the rucio tutorials in particular (https://www.hep.lu.se/courses/grid/2019/NTF004F-lecture6.pdf) and grabbed some stuff from here (https://gitlab.cern.ch/felopez/rucio/-/tree/0.3.5/tools). `pyopenssl.py` comes from https://github.com/urllib3/urllib3/blob/main/src/urllib3/contrib/pyopenssl.py (for python2 in some cases and might be droppable with...

> Also, I would recommend not exposing any metrics outside of your network. Thanks! (For reference to anyone looking at this issue, since it took a little bit of time...

I've managed to do this with a custom `Dockerfile` with some quick `sed` work that seems to work well (using traefik as a reverse-proxy): ``` $ cat multivisor/Dockerfile FROM python:3...

This is a working list of packages, the latest `Flask` (2.2.2) is also incompatible with this downgraded version of `werkzeug`: ``` # python -m pip list Package Version -------------- -----------...

This bit me a bit, and I'm struggling to understand how this sort of interacts with each other (especially in editable installs). It would be great to have clearer documentation...

In `pyhf` where we have histograms in two possible paths (because of how HistFactory works) - avoiding the `DeserializationError` by comparing against a (cached) set of keys gives us a...

So nicely, #209 has a TEfficiency which looks like a good vector of attack (pun intended) for memberwise serialization. This can be easily made like so ```python import ROOT fp...

Here's the full sequence of that TEfficiency with the following histograms stored: ```bash $ cat tefficiency.py import ROOT #fp = ROOT.TFile.Open('uproot4-issue209.root') fp = ROOT.TFile.Open('test-efficiency.root') eff = fp.TEfficiencyName num = eff.GetPassedHistogram()...

Continuing further using this python, it seems to be that the memberwise is complaining about the `AsVector::read` call. So if we look at the streamer for `TEfficiency` here and the...

Ok, getting further, we have a slight issue that I think `uproot` might need to be refactored since I think memberwise streaming breaks the current `Model.read` assumptions. Let me step...