Say you run revdepcheck to just find a small bug, which you then fix. When you then run revdepcheck again a few moments later, it'll rerun both 'old' and 'new' checks, although the 'old' ones will produce identical R CMD check output as the first run.
I'd like to propose a mechanism for revdep_check() to keep existing 'old' results and just rerun the 'new' ones.
For this to be solid, I guess it needs to make sure that none of the revdep packages and none of their dependencies have been updated since the previous run. If they have, then those cached results should be invalidated triggering a rerun of 'old' for the invalidated revdep packages.
The above strategy should decrease the processing time significantly, even if there are a few updates in the revdep DAG.
Say you run revdepcheck to just find a small bug, which you then fix. When you then run revdepcheck again a few moments later, it'll rerun both 'old' and 'new' checks, although the 'old' ones will produce identical
R CMD checkoutput as the first run.I'd like to propose a mechanism for
revdep_check()to keep existing 'old' results and just rerun the 'new' ones.For this to be solid, I guess it needs to make sure that none of the revdep packages and none of their dependencies have been updated since the previous run. If they have, then those cached results should be invalidated triggering a rerun of 'old' for the invalidated revdep packages.
The above strategy should decrease the processing time significantly, even if there are a few updates in the revdep DAG.