Closed HelenaLC closed 11 months ago
outstanding pain points:
- rather than
attach.R
andzzz.R
, would depending (instead of importing) packages be an option? how to other tidy packages deal with defining for S3 methods for foreign generics?
I took those files from the tidyverse package. Would add them to depend load those package analogously to what attach.R does?
Maybe changing this could be another pull request. I am never against simplifying stuff ;)
dplyr
'sbind_rows/cols()
masks thettservice
version; currently need to writettservice::bind_rows/cols()
for methods to work
Yes I'm not sure why bind_*
have not been developed as methods. I would prefer using tidySingleCellExperiment::
because ttservice
should be completely hidden from the user.
If only tidySingleCellExperiment
is loaded or loaded last, this is not a problem. However, I understand that this explicit declaration makes the example more robust.
the good news:
- R CMD check is currently clean
devtools::document()
is warning-free
Amazing!
For 1. I think this is could be due to how ttservice documents them, but will doublecheck. I.e., we get what we inherit, so fixing it upstream might resolve it.
2.-4. Agreed & I can fix those. Re depending vs. attaching - I am not sure. I‘ll try to give it a shirt and see first hand. Plus do some research. Let‘s see!
Re bind_rows etc., still worth thinking about how we can resolve this. Eg, as a user, I get pretty confused with non-informative errors because some other method was dispatched :/
For 1. I think this is could be due to how ttservice documents them, but will doublecheck. I.e., we get what we inherit, so fixing it upstream might resolve it.
Let's think about this after this PR then.
2.-4. Agreed & I can fix those.
When these are done, most likely the PR is good to go!
Looking forward to accept it.
(also please have a look if they pass the github actions, before requiring code review)
Hm. I wonder if these fails could be related to older R/Bioc versions in the GHA? I could not reproduce them on R 4.3/Bioc 3.17, so am a bit stuck how to fix this.
Hm. I wonder if these fails could be related to older R/Bioc versions in the GHA? I could not reproduce them on R 4.3/Bioc 3.17, so am a bit stuck how to fix this.
I see just one warning
Warning: declared S3 method 'rename.SingleCellExperiment' not found
must be some little bug, I don't think it depends on versions of R/Bioc
Hm. I wonder if these fails could be related to older R/Bioc versions in the GHA? I could not reproduce them on R 4.3/Bioc 3.17, so am a bit stuck how to fix this.
I see just one warning
Warning: declared S3 method 'rename.SingleCellExperiment' not found
must be some little bug, I don't think it depends on versions of R/Bioc
Yeah, I think you're right - hoping this'll fix it! Maybe worth considering to eventually up GHA versions in any case, just to be sure we're testing in the same realm as Bioc servers will (as to not get unexpected build/check warnings/errors when pushing upstream).
Hm. I wonder if these fails could be related to older R/Bioc versions in the GHA? I could not reproduce them on R 4.3/Bioc 3.17, so am a bit stuck how to fix this.
I see just one warning
Warning: declared S3 method 'rename.SingleCellExperiment' not found
must be some little bug, I don't think it depends on versions of R/Bioc
Yeah, I think you're right - hoping this'll fix it! Maybe worth considering to eventually up GHA versions in any case, just to be sure we're testing in the same realm as Bioc servers will (as to not get unexpected build/check warnings/errors when pushing upstream).
To have quicker iterations you could even PR onto your master, test, and then push your master into mine after it passes the tests.
The warning is because it thinks, for example, we are extending stats::filter
rather than dplyr::filter
Replaced by stemangiola/tidySingleCellExperiment#75
BiocStyle::CRAN/Biocpkg()
and friendsecosystems (tidyverse), classes/variables (
tibble
), functions (unnest()
)tidySingleCellExperiment
's functionalitydplyr
,tidyr
, ..., methodsattach.R
andzzz.R
from stemangiola/tidySingleCellExperiment#69 to assure needed dependencies are attachedattach.R
andzzz.R
, would depending (instead of importing) packages be an option? how to other tidy packages deal with defining for S3 methods for foreign generics?dplyr
'sbind_rows/cols()
masks thettservice
version; currentlyneed to write
ttservice::bind_rows/cols()
for methods to workdevtools::document()
is warning-free