Note: you might need to add the BioJulia registry. I asked on Slack last month, and David Varela said
grep it says MicrobiomePlots is the offender
I’m not quite sure what he’s grepping - I don’t see MicrobiomePlots in the stacktrace, but the MicrobiomePlots [compat] section does have Microbiome = "0.4, 0.5". I don’t understand why that doesn’t work though.
How is this registry being created? I’m fairly certain it is using an incorrect format for the registry compat files. It should be using a vector of strings where each string is a version range.
The registry is created and updated using an installation of Registrator that makes PRs to the registry.
And the bot registered MicrobiomePlots.
My first thought was ah, perhaps the format has changed, and the BioJulia bot setup needs updating.
Second thought was why we haven’t had that problem before if that’s the case… unless it’s because the release PR was made by an older bot, and then if it wasn’t merged for a while and so the problem wasn’t seen.
This may be related:
Which puts me in mind of the bug in the automerge checker for the general registry when a patch release is made - it would throw a tizwas claiming julia versions were being restricted when they are not.
At some point during the holiday period I was going to replicate General’s CI setups on BioJuliaRegistry so as we can have some more systematic checking… Then I got ill and have just started to recover, so now I intend to look into doing that tomorrow. I’m also going to be releasing a 0.1 of GenomeGraphs to the BioJuliaRegistry soon. Which might give me the opportunity to test the bot’s behaviour by doing a dummy release where GenomeGraphs required Microbiome 0.4 and 0.5 like MicrobiomePlots does, and then I can see what the bot does vs what it should do, then I can just rebuild the bot from a more recent Registrator.jl state if it’s wrong and see if that fixes it.
But since you just want this particular offending entry fixed for your resonance project, if you want to just change the offending line in a PR and I’ll merge it.
Weird… i wonder if I have something else going on triggers something different for the resolver. Frankly, I don’t really understand how all of this works.
You mean a PR direct to the registry? I can also drop support for Microbiome 0.4, perhaps I should do that in any case.