ETA on switch to BioC 3.16
PeteHaitch opened this issue · comments
Hi Dirk,
BioC 3.16 is hot off the presses (https://bioconductor.org/news/bioc_3_16_release/).
Will coverage of BioC by r2u switch or extend to BioC 3.16?
I've no idea what's involved in making such a change, but, if feasible, do you have an ETA?
Thanks,
Pete
It's a good question. I have been pondering it myself for a while with 3.16 on the horizon.
r2u lives and dies by what rspm gives it, and compiles as needed. For BioC, we always compiled. So that is some 200+ packages. Not insurmountable, but some work. I was reminder how to compute dependency depth. So what we need now, I think, is a simple routine that
- loops over all BioC packages in the r2u universe
- computes their 'full' dependency length
- loops over these from the inside out:
- dependency count 0 (BiocManager likely?)
- dependency count 1
- ....
- dependency count N (for whatever value of N)
Which seems feasible. So ... can you help? I am still a wee bit tied up as it is a teaching term for me.
So ... can you help? I am still a wee bit tied up as it is a teaching term for me.
Not immediately, unfortunately.
Next week I'm examining Masters theses and doing final organisation of BioC Asia for end of the month.
I can try to return to it in ~2 weeks and see where I can get with it
A few days ago I noodled around with the package dependency 'graph' and worked out a way to script this. With some time today I ran it for 'jammy' aka 22.04 -- and the BioConductor packages included in that half of r2u have now been rebuilt for BioConductor 3.16. And I just started 'focal' aka 20.04 for completeness.
This is now complete.
Amazing, thanks Dirk!
Let me know if you find particular (popular) packages missing. The BioC side is more manual than the CRAN side so I do not think I can get to full coverage (== venue for volunteers to shine ...) but I could add a few if it made sense. What we have now is determined strictly by what CRAN pulls is -- so a wee bit random.