Open yarikoptic opened 5 years ago
this causes lots of grief... even can't connect via detached debugger (epdb):
$ asv run -j 4 -e -b api.supers -E existing
...
File "/home/yoh/proj/datalad/datalad/benchmarks/api.py", line 88, in setup
import epdb; epdb.serve()
File "/home/yoh/proj/datalad/datalad/venvs/dev/local/lib/python2.7/site-packages/epdb/__init__.py", line 1084, in serve
Epdb().serve(port)
File "/home/yoh/proj/datalad/datalad/venvs/dev/local/lib/python2.7/site-packages/epdb/__init__.py", line 150, in serve
Epdb._server.handle_request()
File "/home/yoh/proj/datalad/datalad/venvs/dev/local/lib/python2.7/site-packages/epdb/epdb_server.py", line 238, in handle_request
self._serve_process(err.slaveFd, err.serverPid)
File "/home/yoh/proj/datalad/datalad/venvs/dev/local/lib/python2.7/site-packages/epdb/epdb_server.py", line 251, in _serve_process
if sys.stdin.isatty():
ValueError: I/O operation on closed file
so question remains - if intentional or not (then should be fixed)
somewhere between v0.2.2 and v0.3b1 asv started to cause stdin being closed in the ran benchmarks, so our suit started to fail with e.g.
I wonder if such change in behavior was intentional (which is IMHO a bit too aggressive) or may be just a side-effect?