[Discuss] Weird awk processing

[gaf at blu.org (Jerry Feldman)]

On 05/02/2013 07:17 PM, David Rosenstrauch wrote:
> On 05/02/2013 03:52 PM, Jerry Feldman wrote:
>> On 05/02/2013 01:11 AM, David Rosenstrauch wrote:
>>> Just stumbled upon the most bizarre awk problem.  mawk and gawk are
>>> showing 2 different results for the same code.  Can anyone shed any
>>> light?
>>>
>>> TIA!
>>>
>>> DR
>
>> I get the same results from the 2. The first thing I did was to run
>> sample.txt through sed to convert the spaces to tabs. Certainly the
>> correct result is the gawk result you got.
>
> The issue apparently only occurred in certain versions of mawk. (Most 
> notably v1.3.3, which was installed onto my Ubuntu server.)
>
> I (with some help from others) managed to pinpoint the issue earlier 
> today, fortunately:  apparently mawk was interpreting "32e56067" and 
> "32e56122" as numbers coded in scientific notation, interpreting them 
> both as infinity, and then deciding they were equal!  The fix was to 
> change this line:
>
> curr_label=$1;
>
> to this:
>
> curr_label=""$1;
>
> to force mawk to interpret $1 as strings.  Go figure!
>
Blame is on Mark Shuttleworth :-)

-- 
Jerry Feldman <gaf at blu.org>
Boston Linux and Unix
PGP key id:3BC1EB90
PGP Key fingerprint: 49E2 C52A FC5A A31F 8D66  C0AF 7CEA 30FC 3BC1 EB90