On Sat, Jun 27, 2020 at 09:40:44AM -0400, Aisha Tammy wrote: > > I usually try to at least runs through the TEST_DEPENDS consumers and > > check before and after "make test" results. > Is there a fast way of getting this list? > I've only got the basics of make regress for a port and how to do it > for multiple subdirs, subdirs="cat1/port1 cat2/port2" make regress. > Is there a better way?
I use danj's showvictims.py script: https://github.com/danieljakots/obsd-ports-tools It queries the sqlports sqlite database. Other folks just query that directly. I usually do them by hand since I need to see the results. It's not uncommon that not all tests pass for python ports, so I usually end up comparing before and after the port update. (No reason to be super worried if one's port update makes *less* tests fail than the original). > Couple of more questions: > If some packages are failing because they are not updated, would I > need to update them in this same patch? I usually send them separately but note that it may be needed for an update. Sometimes those updates will work just fine with the old version of the original port. I prefer one port per email myself, just with notes about "needs version X.X of py-foo that I sent previously" > If there is no update to them, would the consensus be to keep old packages > instead of updating this one? If your update breaks a number of other ports, we'd tend to keep the old version rather than break other ports. --Kurt
