[Python-Dev] Unary minus bug (original) (raw)
Neal Norwitz nnorwitz at gmail.com
Mon Jul 10 00:23:26 CEST 2006
- Previous message: [Python-Dev] Unary minus bug
- Next message: [Python-Dev] xml issue in 2.5
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]
On 7/9/06, Neil Schemenauer <nas at arctrix.com> wrote:
On Sun, Jul 09, 2006 at 03:02:06PM -0700, Neal Norwitz wrote: > Do we care about this (after your checkin and with my fix to make > 32-63 bit values ints rather than longs): > > # 64 bit box > >>>minint = str(-sys.maxint - 1) > >>>minint > '-9223372036854775808' > >>>eval(minint) > -9223372036854775808 > >>>eval('-(%s)' % minint[1:]) > -9223372036854775808L
I don't think we care. Python <= 2.4 has the same behavior (i.e. adding parens defeats the constant folding since it's done at the syntax tree level).
Ok, unless someone care's enough to make a patch, I'm removing this from the outstanding issues for 2.5.
Where's your fix regarding 32-63 bit ints? I'm not familiar with
I just had to finish the test. It's checked in now and should fix the buildbot problems. Your test exposed the issue, but didn't cause it. It was caused by the optimization of converting strings to ints. My approach to fix it was kinda hacky (just add the proper table for 64-bit machines). I guess ideally we would generate this table based off sizeof(long). Probably should have added a TODO.
n
- Previous message: [Python-Dev] Unary minus bug
- Next message: [Python-Dev] xml issue in 2.5
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]