Check table_path() approach with backends

[hadley]

And figure out what fixes are needed.

[hadley]

• bigrquery: https://github.com/r-dbi/bigrquery/pull/555/files
• odbc: nothing needed
• pool: nothing needed (I was thinking of rstudio/pool@8ce7c0b, but that seems to a pool infelicity not primarily a dbplyr breaking change)

[hadley]

revdep failures that look related:

• diseasystore — from SCDB
• pool: needs sql_escape_ident method. Fixed in Switch methods to default, not DBIConnection #1467
• SCDB — dbplyr 2.5.0 compatibility ssi-dk/SCDB#114

Also need to work through packages that use dbplyr_table_ident:

• bigrquery: Updates for dbplyr 2.5.0 r-dbi/bigrquery#601
• editbl — should be ok, as fallback for non-dbplyr_table_ident looks ok.
• implyr — dbplyr 2.5.0 compatibility ianmcook/implyr#62
• ramses — Forward compatibility for dbplyr 2.5.0 ramses-antibiotics/ramses-package#117
• RPresto — Forward compatibility with dbplyr 2.5.0 prestodb/RPresto#265

[hadley]

cc @mgirlich can you think of any other ways to find packages that might be affected? Otherwise I now feel pretty confident that the existing API is adequate and we can close this issue.

[mgirlich]

I also looked at the PRs you created and I agree that the exported API looks good. For finding affected packages I also relied on revdep, so I don't know any other way to find affected packages.