We (@FelixGuenther, @medewitt, and myself) had a fairly casual chat today. We covered:
- Our new seminar series with the broad consensus being that we liked it and were looking forward to the scheduled talks. @FelixGuenther is finalising the speaker for March and will have an announcement post soon.
- We chatted about improving the community meetings and agreed that more technical/specific discussion should be left for the second half and if this starts overrunning we can think about a separate meeting. I agreed again to be better at announcing meetings.
- We had a brief discussion of issues using
rstanand its development time line, issues shipping tools that rely oncmdstan, and the promise ofJulia(and potentially how it may not live up to that promise when tested in anger). - We had a nice chat about linking model modules or joint modelling and agreed it often comes down to documentation. We discussed how nice a meta-map of the kinds of questions that might be asked, the kind of data that might be available, and the kind of approaches that might be taken to get from data to answers would be helpful in designing our tools and documenting how to use what we currently have. @medewitt suggested we could start putting something together using https://app.diagrams.net/. I really think this could be incredibly useful so if anyone is interested let’s chat.
- We talked through @medewitt data on a local Mpox outbreak (see Handling delayed entry of symptom onset dates in line lists - #10 by samabbott) and how it related to current tools. We floated the idea of a case report-like document detailing what kind of data people have, what kind of questions they need to ask, and how these both change over time. Potentially we could use a community seminar to bring multiple people with this kind of applied expertise together so we can tease out common threads.
- We also chatted about how visualisation and data formats are a big blocker to understanding data and what we can do about this. @FelixGuenther has a branch with some nice visualisations (see Add observations visualisations and vignette · Issue #23 · epinowcast/epinowcast · GitHub) which we can hopefully better surface soon. Joel Hellewell also started some work mapping between formats that we could look at wrapping up (see Add data format converters · Issue #144 · epinowcast/epinowcast · GitHub).
- Lastly, we talked about
{touchstone}a cool new package for benchmarking code in CI and how this would be nice forepinowcast. Thanks to @sbfnk we have a working implementation of this in{EpiNow2}(EpiNow2/touchstone at main · epiforecasts/EpiNow2 · GitHub) and it has already proved super helpful in detecting performance regressions (implement `dist_spec` interface by sbfnk · Pull Request #363 · epiforecasts/EpiNow2 · GitHub)
The big news is that @medewitt taught me about font ligatures in vscode so now my code looks like this:
which I am sure we can all agree is amazing.
To do this see How to Enable Font Ligatures in VS Code (or I guess lots of other places on the interwebs)
If you want to have a similarly dreamy time in Rstudio then this guide (Using Fira Code Ligatures in RStudio – Jeffrey Girard) looks good (though do think about vscode it is like living in the dream (i.e. Inception (2010) Official Trailer #1 - Christopher Nolan Movie HD - YouTube))