Bug in optimization of range

[robertwb]

This:

   cdef int i, n
    n = 10
    for i in range(n):
        if i == 5: n *= 2
    print i

prints out 20.

Migrated from http://trac.cython.org/ticket/203

[robertwb]

@dagss commented

It should print 9, I think

[robertwb]

@dagss commented

Note: This must be changed in the for in range-optimization, because "for 0 <= i < n" should have this semantics!

[robertwb]

@dagss changed milestone from 0.11 to 0.11.1
commented

[robertwb]

@dagss changed owner from somebody to Prajwal Suhas P
commented

[robertwb]

@robertwb changed resolution to fixed
status from new to closed
commented

Works as expected. Committed http://hg.cython.org/cython-devel/rev/620326b14c3f

This is another reason to keep the for...from syntax though.

[robertwb]

@robertwb changed component from Optimization to Python Semantics
commented

[robertwb]

@robertwb commented

This is a dupe of http://trac.cython.org/ticket/196, the above commit was for http://trac.cython.org/ticket/208.

[robertwb]

@dagss changed resolution from fixed to empty
status from closed to reopened
commented

No, this is a different bug. Reopening.

Also note that this ticket is assigned to Prajwal Suhas P; see http://thread.gmane.org/gmane.comp.python.cython.devel/4775

Magnus' patch was only for incrementing-one-to-far; this is about evaluating the endpoints to range once per loop.

Also consider this test case:

"""
 >>> f()
g called
0
1
2
2
"""

cdef int g():
     print "g called"
     return 3

def f():
     cdef int i
     for i in range(g()):
         print i
     print i

[robertwb]

@dagss commented

That is; I get:

Failed example:
    f()
Expected:
    g called
    0
    1
    2
    2
Got:
    g called
    0
    g called
    1
    g called
    2
    g called
    2

[robertwb]

@dagss changed priority from major to critical
commented

[robertwb]

@robertwb changed resolution to fixed
status from reopened to closed
commented

Finally resolved by http://trac.cython.org/ticket/243