[tahoe-lafs-trac-stream] [tahoe-lafs] #1418: "cannot convert float NaN to integer" in next_power_of_k, during upload via helper

[tahoe-lafs]

#1418: "cannot convert float NaN to integer" in next_power_of_k, during upload via
helper
------------------------+---------------------------
     Reporter:  rycee   |      Owner:  rycee
         Type:  defect  |     Status:  new
     Priority:  major   |  Milestone:  undecided
    Component:  code    |    Version:  1.8.2
   Resolution:          |   Keywords:  helper upload
Launchpad Bug:          |
------------------------+---------------------------

Comment (by zooko):

 It can't be hexed hardware, because it is reproducible (at least mostly?)
 and it doesn't happen with any other executables, nor with other Python
 experiments. Say, here's an idea. My best guess is that one of the native-
 code dependencies (hopefully not zfec or pycryptopp -- the ones I'm
 responsible for!) are doing something bad. You could test this by running
 some complicated floating point math, then importing each native code
 dependency (see comment:23) and then running the floating point math
 again.

 Also the tracing approach that I suggested in comment:24 is quite likely
 to give us useful information. If you're not comfortable with all the
 Python coding suggested on http://www.doughellmann.com/PyMOTW/trace/ then
 I can whip up a script that'll do it for you. :-)

-- 
Ticket URL: <http://tahoe-lafs.org/trac/tahoe-lafs/ticket/1418#comment:26>
tahoe-lafs <http://tahoe-lafs.org>
secure decentralized storage