On Wed Aug 03 13:36:01 2005, wfischer@lanl.gov wrote: > Dprof/dprofpp reports incorrect subroutine call count if calls > 999999 > (possibly the problem is in dprofpp) > > In this version of Dprof/dprofpp, the #Calls field is apparently > limited to > 6 characters: values >= 10**6 are truncated (keeping the left-most 6 > digits only). > > For example, the following one-liner calls a no-op subroutine 1,234,567 > times, > but only 123,456 calls are reported: > Problem confirmed. Can we assume that setting an upper bound to the number of calls is necessary and prudent? If so, what would be more rational behavior? Running 999,999 calls and then issuing a warning? Are there other versions of dprofpp which behave differently/better? Thank you very much. Jim KeenanThread Previous