[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):

 Replying to [comment:22 rycee]:
 > I have previously used 1.8.1 reliably as a helper so I suppose the
 problem might be caused due to some newer version of a library pulled in
 during `python setup.py build`?

 Yes, Tahoe-LAFS itself is written in pure Python. I suspect this problem
 is due to memory corruption or some other incorrect C/C++ code, so it is
 probably due to one of the dependencies. I really want to track down which
 one it is, so please try to continue maintaining a "paper trail" of what
 you changed at each step of this. Thanks!

 To see which dependencies have native code in them, look at this table:
 http://tahoe-lafs.org/source/tahoe-lafs/deps/tahoe-lafs-dep-
 eggs/README.html Every column there is a library that uses native (C/C++)
 code.

 If I were you, my next step would be to clear the false positives out of
 valgrind (as I described in comment:18). valgrind will either identify the
 problem, or if it gives a clean report then we can be confident that there
 is ''no'' branching on uninitialized memory, writing to memory out of
 bounds, or other of the memory errors that valgrind checks for.

 If you're not interested in doing that then say so and lets think of some
 other approaches to track down this bug.

-- 
Ticket URL: <http://tahoe-lafs.org/trac/tahoe-lafs/ticket/1418#comment:23>
tahoe-lafs <http://tahoe-lafs.org>
secure decentralized storage