Closed jordansread closed 4 years ago
Cool, Jeff suggested something similar this week that you may be referring to (https://github.com/jsadler2/ds-pipelines-3/pull/11#issuecomment-657057885).
I thought this would be extremely quick, but some silly decisions about how to treat 1 (and 0?) files in sc_indicate
came to roost when I tried to write combine_to_tibble
the simple way.
Example for some temp files: for multiple files, we get list names = filenames and list elements = hashes:
> sc_indicate('', data_file=tfiles[1:3])
$`/var/folders/_0/fbg0ffkj6z3fb_7jvm8wg86r002c5h/T//RtmpDIhLsH/139e03beea864`
[1] "873fabd21220be3cef08dd074dbd8f19"
$`/var/folders/_0/fbg0ffkj6z3fb_7jvm8wg86r002c5h/T//RtmpDIhLsH/239e031715bf9`
[1] "d41d8cd98f00b204e9800998ecf8427e"
$`/var/folders/_0/fbg0ffkj6z3fb_7jvm8wg86r002c5h/T//RtmpDIhLsH/339e094d51cb`
[1] "aa09a0b52ac9414d807e3a13275cdd3d"
but for just one file we get a list where the name is hash
instead of the filename:
> sc_indicate('', data_file=tfiles[1])
$hash
[1] "873fabd21220be3cef08dd074dbd8f19"
Should we change the behavior of sc_indicate
or write around it with combine_to_tibble
(and also combine_to_ind
?)?
I'm leaning toward working around because changing it would potentially cause a lot of unnecessary rebuilds in existing projects. It feels untidy, though.
combine_to_ind(ind_file, ...)
returns a file. But there are similar cases where you want to do the same thing but return an R object (e.g., a tibble withfilepath
andhash
columns).perhaps something like
combine_to_tibble(...)
(or_to_table
to avoid the tibble/data.frame naming)?I thought maybe this would be what
coming_to_ind
could do if you specifyind_file = NULL
, but that wouldn't work with the current version of combiner functions, which don't have much flexibility for arguments.