Hi!
I've put together a tiny patch for the for loop bug, and just thought
I'd run it by you guys. As far as I understand, the "for ... from"
node is also used for optimized range()-loops (i.e., over C integer
indices) -- so that's the only place I fixed anything. (It seems to
work for range too, now.)
I've basically just mirrored the incop, creating an inverse named
decop, to "undo" the last (superfluous) step, and then put a statement
using this (such as "i--" if the incop is "i++", or "i-=2" if the
incop is "i+=2") right after the closing parens of the for loop. As
far as I can see, this should be safe, because if the code breaks out
of the loop (using a goto in C), the last step will never occur, so
the "undo" isn't needed.
--- a/Cython/Compiler/Nodes.py Wed Dec 17 01:43:58 2008 -0800
+++ b/Cython/Compiler/Nodes.py Fri Jan 30 12:02:15 2009 +0100
@@ -3764,9 +3764,12 @@
self.bound1.generate_evaluation_code(code)
self.bound2.generate_evaluation_code(code)
offset, incop = self.relation_table[self.relation1]
+ decop = "++" if incop == "--" else "--"
if self.step is not None:
+ step = self.step.result()
self.step.generate_evaluation_code(code)
- incop = "%s=%s" % (incop[0], self.step.result())
+ incop = "%s=%s" % (incop[0], step)
+ decop = "%s=%s" % (decop[0], step)
code.putln(
"for (%s = %s%s; %s %s %s; %s%s) {" % (
self.loopvar_name,
@@ -3778,7 +3781,7 @@
self.target.generate_assignment_code(self.py_loopvar_node, code)
self.body.generate_execution_code(code)
code.put_label(code.continue_label)
- code.putln("}")
+ code.putln("} %s%s;" % (self.loopvar_name, decop))
break_label = code.break_label
code.set_loop_labels(old_loop_labels)
if self.else_clause:
I've basically just tested it with loops from 0 to 10 and from 10 to
0, with steps 1 and 2, with both loop syntaxes (for ... from and
for ... in range(...)).
Of course it's very possible that I've missed something obvious
somewhere... :->
If it seems OK, I could always write up some more proper tests and
upload a patch including those to Trac for review.
--
Magnus Lie Hetland
http://hetland.org
_______________________________________________
Cython-dev mailing list
[email protected]
http://codespeak.net/mailman/listinfo/cython-dev
