Raymond Hettinger [EMAIL PROTECTED] added the comment:
Recommend rejecting. Too much work for zero performance payoff.
--
assignee: rhettinger -
___
Python tracker [EMAIL PROTECTED]
http://bugs.python.org/issue1394
Guido van Rossum [EMAIL PROTECTED] added the comment:
Rejecting. Too much risk as well, based on experience in the past with
supposedly harmless optimizations.
--
resolution: - rejected
status: open - closed
___
Python tracker [EMAIL PROTECTED]
Alexander Belopolsky added the comment:
I have added more tests as Neal suggested. See unreachable-code-with-
tests.diff for a combined patch. I've rerun all tests after removing
.pyc files and also with -O option (ensuring no existing .pyo files as
well). All works.
I've removed #if 0
Paul Pogonyshev added the comment:
Well, I cannot guarantee it will _always_ be eliminated, since I too
don't really know when generated bytecode can (currently) end in
non-return. However, if before it ended in non-return, that non-return
must have been unreachable, obviously (else code flow
Paul Pogonyshev added the comment:
Actually, it is even better. Since you don't increment codelen, that
extra RETURN_VALUE is virtual in that it acts as a sentinel, but will
not appear in the resulting bytecode. You can test this by backing out
the patch and disassembling something. For this
Alexander Belopolsky added the comment:
Since you don't increment codelen ..
Good catch!
For this reason, if() before
writing the RETURN_VALUE is not needed.
In this case it will be clearer to use STOP_CODE instead of RETURN_VALUE
as a sentinel.
__
Tracker
Paul Pogonyshev added the comment:
Yes, help with unit tests would be appreciated. Especially since it is
not supposed to fix anything, so I'm not sure what unit tests should be
like...
BTW, trying to understand why the patch didn't remove unreachable code
in your first example, I noticed this
Paul Pogonyshev added the comment:
Speaking of which, I propose that codestr[] array is made one byte
longer and RETURN_VALUE opcode wrote in that extra byte. It will be
removed by this patch anyway (if it is accepted), but we'll remove a way
to accidentally disable peephole optimizer.
I mean,
Benjamin Peterson added the comment:
Yes, help with unit tests would be appreciated. Especially since it is
not supposed to fix anything, so I'm not sure what unit tests should be
like...
Unit tests are just for bugfixes. They let us make sure Python is doing
what we want it to do in a given
Guido van Rossum added the comment:
On Mon, Feb 25, 2008 at 1:32 PM, Benjamin Peterson
[EMAIL PROTECTED] wrote:
Unit tests are just for bugfixes.
I presume you meant are *not* just for bugfixes.
__
Tracker [EMAIL PROTECTED]
http://bugs.python.org/issue1394
Benjamin Peterson added the comment:
I presume you meant are *not* just for bugfixes.
(Sigh) Yes, of course.
__
Tracker [EMAIL PROTECTED]
http://bugs.python.org/issue1394
__
___
Changes by Raymond Hettinger:
--
assignee: - rhettinger
nosy: +rhettinger
__
Tracker [EMAIL PROTECTED]
http://bugs.python.org/issue1394
__
___
Python-bugs-list mailing list
Paul Pogonyshev added the comment:
Thanks for writing the test.
Yes, I did read the comment. As I understood it, RETURN_VALUE is needed
only so that various optimization can assume codestr[] cannot suddenly
end without one. E.g. if you match for BINARY_ADD, you can safely check
the next
Alexander Belopolsky added the comment:
Paul,
You are right. I misunderstood that comment myself. I've tried the
attached modification to your patch (unreachable-code-1.diff) and it
passes all tests while fixing msg62953 problem:
$ cat t.py
def f():
return 1
1+2
from dis import
Alexander Belopolsky added the comment:
I am not sure what this patch would accomplish. I tried
$ cat t.py
def f():
return 1
1+2
from dis import dis
print dis(f)
Both with and without patch I get
$ ./python.exe -O t.py
2 0 LOAD_CONST 1 (1)
3
Neal Norwitz added the comment:
It would be great to see test cases with this change. That would help
answer Alexander's question too.
--
nosy: +nnorwitz
__
Tracker [EMAIL PROTECTED]
http://bugs.python.org/issue1394
__
Changes by Guido van Rossum:
--
nosy: +gvanrossum
__
Tracker [EMAIL PROTECTED]
http://bugs.python.org/issue1394
__
___
Python-bugs-list mailing list
Unsubscribe:
New submission from Paul Pogonyshev:
This patch improves bytecode output, by removing unreachable code. It
doesn't target special cases, as now, but provides a generic implementation.
--
components: Interpreter Core
files: unreachable-code.diff
messages: 57141
nosy: Paul Pogonyshev
18 matches
Mail list logo
