alarm() sometimes too short in rm_create.t?

[eserte]

The five fail reports listed on
http://analysis.cpantesters.org/reports_by_field?distv=Filesys-Notify-Simple-0.12;field=fail%3At%2Frm_create.t (state=FAIL and fail:t/rm_create.t=0) look like the alarm() period was too short here; the test script runs here were killed with SIGALRM.

[miyagawa]

Thanks. The analysis is correct but I'm not sure increasing the alarm might help - it might be just in a race condition or some sort.