sprhawk / python-on-a-chip

Automatically exported from code.google.com/p/python-on-a-chip
Other
0 stars 0 forks source link

Make generators work in ipm #95

Closed GoogleCodeExporter closed 9 years ago

GoogleCodeExporter commented 9 years ago
Report from ziglionz on maillist (2010/06/27)

At ipm prompt:

def gen(o):
   yield o

def d(o):
  for i in gen(o):
     print i

d(123)

it just works once then PyMite crashes

Original issue reported on code.google.com by dwhall...@gmail.com on 27 Jun 2010 at 11:53

GoogleCodeExporter commented 9 years ago
Some more observations.
Interestingly, I'm seeing the same problem running the code not from ipm but 
rather as a user module.
Have added some debug code:

def gen(o):
    print "enter gen"
    yield o
    print "ret gen"

def d(o):
   for i in gen(o):
       print i
   print "ret d"

After calling d(123), the screen output is:

enter gen
123
ret gen

So I never saw the calling function d() returning. It still looks like a crash.
E.

Original comment by zigli...@gmail.com on 28 Jun 2010 at 12:58

GoogleCodeExporter commented 9 years ago
Simpler case fails in ipm:

ipm> def gen(o):
 yield o

ipm> for i in gen(4):
 print i

 4 
ipm> a=4
Connection write error, type Ctrl+D to quit.
Connection read error, type Ctrl+D to quit.

Original comment by dwhall...@gmail.com on 5 Aug 2010 at 3:12

GoogleCodeExporter commented 9 years ago
Problem found!

The code at the very bottom of the VM's for() loop that was supposed to find a 
B_LOOP block when a StopIteration exception occurs was not exiting its search 
loop when the block was found.

Non-interactive mode didn't display this defect because when the StopIteration 
exception was raised in the test code, there was only one loop structure around 
the executing code.  Whereas, in interactive mode, src/lib/ipm.py has a "while 
1" loop around the read-eval-print loop that runs ipm.  So in interactive mode, 
the VM was finding the B_LOOP block of that outer while loop and using it, 
which led to ipm and the VM completing code execution and exiting.

Original comment by dwhall...@gmail.com on 7 Aug 2010 at 3:54

GoogleCodeExporter commented 9 years ago
r566
- Removed code in RETURN_VALUE bytecode that tried to handle StopIteration 
early.
- Fixed the StopIteration exception handler at the bottom of the VM bytecode 
loop.  The "break" statement was only breaking out of the innermost "while" 
loop.
- Added an "assert" line to ipm to catch when a real StopIteration gets caught 
by ipm's "while 1" loop.

All tests pass, including manual ipm tests in the comments above.

Original comment by dwhall...@gmail.com on 7 Aug 2010 at 4:10

GoogleCodeExporter commented 9 years ago
Excellent, I don't know much about the internals but I'm glad ipm behaves even 
more like the non-interactive mode.

Original comment by zigli...@gmail.com on 7 Aug 2010 at 8:36