[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: optimal-page-breaking: signed/unsigned bug; issue 1553 (issue 227500
|
From: |
Keith OHara |
|
Subject: |
Re: optimal-page-breaking: signed/unsigned bug; issue 1553 (issue 22750045) |
|
Date: |
Wed, 20 Nov 2013 11:26:59 -0800 |
|
User-agent: |
Opera Mail/12.16 (Win32) |
On Wed, 20 Nov 2013 00:53:24 -0800, <address@hidden> wrote:
If min_sys_count would happen
to be negative, this loop will never get run at all
Exactly that was happening with the example from issue 1553, before the patch.
The code above seemed to be trying to ensured that min_sys_count >= 0 but
thinking in terms of signed arithmetic. Probably better to leave min_sys_count
unsigned and write the subtractions above as saturating subtractions. If you have a
favorite idiom for (x > y) ? x - y : 0 that would be clear up in the changed
code further above, type it in.