After running "rforest," an eclass program available on repec that does not post estimates, a subsequent call to _rmdcoll will fail with the error message "last estimates not found" r(301). For example, if one tries to run "ivregress" after a call to rforest, Stata will produce the error message because ivregress in turn calls _rmdcoll. I suspect the reason is that _rmdcoll attempts to "hold" previous estimates if the macro `e(cmd)' is not blank. This will be the case if an eclass program had been called previously, but if the eclass program that was run did not post estimates, the attempt to hold previous estimates will fail and produce the error message. Probably programs that do not post estimates should not be classified as eclass (but rather rclass or sclass or something) but _rmdcoll should also be more robust. This would be a good improvement.
-
Login or Register
- Log in with
