[Numpy-tickets] [NumPy] #548: fromfile crashes due to failed allocation

[NumPy]

#548: fromfile crashes due to failed allocation
-----------------------------+----------------------------------------------
 Reporter:  albertstrasheim  |       Owner:  somebody     
     Type:  defect           |      Status:  new          
 Priority:  normal           |   Milestone:  1.0.4 Release
Component:  numpy.core       |     Version:  none         
 Severity:  major            |    Keywords:               
-----------------------------+----------------------------------------------
 I'm seeing a crash that appears to be due to fromfile or something it
 calls ignoring a failed allocation.

 Code to reproduce:

 {{{
 import numpy as N

 def main():
     dtype = N.dtype('f4')
     iocount = 6

     x = N.zeros((33554432,), dtype=dtype)
     assert x.nbytes == 128 * 1024 * 1024
     count = len(x)

     print 'writing'
     fp = open('matrix.out', 'wb')
     for i in xrange(iocount):
         x.tofile(fp)
     fp.close()

     del x

     fp = open('matrix.out', 'rb')
     toomuch = []
     for i in xrange(100):
         print i
         fp.seek(0)
         y = N.fromfile(fp, dtype=dtype, count=5*count)
         print len(y)
         toomuch.append(y)
     fp.close()

 if __name__ == '__main__':
     main()
 }}}

 Output on my Windows machine, which can't handle more than one 640 MB
 allocation even though it has 2 GB RAM (heap fragmentation rocks):

 {{{
 writing
 0
 167772160
 1
 167772160 items requested but only 0 read
 }}}

 Crash dialog:

 {{{
 ---------------------------
 python.exe - Application Error
 ---------------------------
 The instruction at "0x10022439" referenced memory at "0x0000001c". The
 memory could not be "read".


 Click on OK to terminate the program
 Click on CANCEL to debug the program
 ---------------------------
 OK   Cancel
 ---------------------------
 }}}

-- 
Ticket URL: <http://projects.scipy.org/scipy/numpy/ticket/548>
NumPy <http://projects.scipy.org/scipy/numpy>
The fundamental package needed for scientific computing with Python.