Open nick-robo opened 2 years ago
I have had similar concerns. But I think it would be helpful to look closely at all available functionality and give a clear suggestion on revision -- even before putting the offered work in -- and the offer is truly appreciated.
Here's an example. The current devel image of GenomicRanges is broken, crashes BiocCheck. But with a small change to one of the vignettes, I can do
> x = BiocCheck::BiocCheck("GenomicRanges")
where R is running in the folder enclosing a checkout of GenomicRanges sources.
Now
> names(x)
[1] "getLastCheck" "show_meta" ".refClassDef" "add"
[5] ".->error" ".->verbose" "error" ".->check"
[9] "check" "initFields" ".self" ".->log"
[13] "verbose" "setCheck" ".->metadata" "zero"
[17] "report" "initialize" "log" "warning"
[21] "metadata" ".->warning" "getBiocCheckDir" ".->note"
[25] "note" "composeReport" "getNum"
> x$error
$checkWatchedTag
$checkWatchedTag[[1]]
[1] "* ERROR: Maintainer must add package name to Watched Tags on the support site; Edit your Support Site User Profile to add Watched Tags."
I don't have to read the output to find the ERROR class and statement.
> names(x$warning)
[1] "checkIndivFileSizes" "checkCodingPractice" "checkForValueSection"
is also useful, and more details can be extracted.
There is also the log file in the GenomicRanges.BiocCheck folder created by BiocCheck.
I am somewhat familiar with the cli API, but I am not completely clear on what aspects of its formatting in, e.g., rcmdcheck, are needed here.
Is there a specific reason why BiocCheck doesn't implement user-friendly outputs, such as
rcmdcheck
anddevtools
, by using thecli
package?
Hi Nick, @nick-robo
The reason is BiocCheck
was developed before the cli
package came around. We would like to add it but it is currently not a priority of the package as it is only cosmetic.
The current devel image of GenomicRanges is broken, crashes BiocCheck.
Hi Vince, @vjcitn
It looks like the Rnw vignette has duplicate chunk names: see https://github.com/Bioconductor/GenomicRanges/pull/72
Unfortunately, BiocCheck
currently does not have a tryCatch
mechanism for these type of errors (as they are few and far between). It would be good to have these error messages recorded in the GenomicRanges.BiocCheck
report.
Currently, BiocCheck outputs a bunch of red text which make it very difficult to read and for users to quickly identify what is wrong and what is working as expected.
Is there a specific reason why BiocCheck doesn't implement user-friendly outputs, such as
rcmdcheck
anddevtools
, by using thecli
package?I would be happy to put some work into this if it's not something that was already considered.