hacker.org

Posted: **Wed Aug 12, 2009 10:06 pm**

Hi,

did you really have access to a system/370?
I considered using an emulator (hercules), but finally managed to convert the floats to their binary representation and that to EBCDIC. Very nice and highly informative challenge!

Posted: **Thu Aug 13, 2009 7:08 pm**

Thanks. I have access to a System z mainframe running z/OS. However, I am not a COBOL programmer. Actually, this little program was my second COBOL program at all. Nevertheless, you described the supposed way to solve this challenge.

Posted: **Thu Sep 17, 2009 1:13 am**

i didnt realize they used a different base for the exponents but wow i had no clue they had a different keyboard scheme... i eventually stumbled onto this little gem here

http://support.microsoft.com/kb/216399/EN-US/

after finding examples to get my excel sheet to work it was solved

Posted: **Tue Sep 29, 2009 10:10 am**

python is your friend again: it has integrated EBCDIC support.

Posted: **Fri Oct 30, 2009 12:12 am**

Very funny challenge (well, up to a certain point)!!
First thing I tried was to download every available free Cobol compiler. But every one of those stumbled over some other dirty detail and produced various errors which I could not fix.
Then I did like you expected... deal with the Cobol syntax, EBCDIC and started doing it all by hand... grrr.
How did you calculate the scientific binary? I was asking Wolfram Alpha, but noticed various rounding errors which resulted in wrong letters here and there. It was always a difference of just 1, so it was easy to find the right letter... but it left a bad smell about trusting floating point values in general...

Posted: **Sun Nov 29, 2009 8:50 am**

Very nice challenge - I had to learn a bit COBOL and IBMs floating point representation before a little Python program could do the decoding.

Posted: **Sat Mar 20, 2010 2:07 pm**

Hi, very nice challenge. and very hard too... I did it all by hand after trying millions of openCOBOLS etc.
http://en.wikipedia.org/wiki/IBM_Floati ... chitecture
http://en.wikipedia.org/wiki/Binary_numeral_system
http://en.wikipedia.org/wiki/Ebcdic
It would be relatively easy without these giant exponents...
Thank you! Really interesting!

Posted: **Tue Sep 07, 2010 12:57 pm**

Really nice challenge.
Figured it out myself, but with some googling around you can find that somebody (polarlemniscate) tried to solve this challenge with help of ibm mainframe user forum: http://www.ibmmainframeforum.com/viewto ... f=6&t=3784
There is answer available in plain text. It would be nice to change float's to say something else.

Posted: **Thu Sep 09, 2010 10:16 pm**

Thanks for the advise. The challenge has been updated.

Posted: **Tue Aug 09, 2011 10:24 am**

z390 for the win.

Posted: **Wed Aug 10, 2011 10:00 am**

AMindForeverVoyaging wrote:z390 for the win.

you mean zCOBOL

I did some research on Cobol and how it represents floats, but i stumbled upon zCOBOL...

Good job to Teebee, very nice challenge.

Posted: **Wed Aug 10, 2011 11:04 am**

contagious wrote:you mean zCOBOL

Whatever the name, as long as I don't have to fiddle with silly things like COMP-1 and EBCDIC.

Posted: **Mon Mar 18, 2013 2:23 pm**

I very much enjoyed the challenge. It introduced me to the Hercules emulator and the z390 / zCobol software. I used the latter one to solve the challenge.

Posted: **Tue Dec 24, 2013 3:17 am**

One of the best challenges, even tricky for an experienced PC COBOL programmer...

Posted: **Fri May 16, 2014 7:47 pm**

I have found this link http://www.ibmmainframeforum.com/viewto ... f=6&t=3784,
what gave me enough info to deduce COMP-1 float representation. rest I did in excell.

hacker.org

Mainframe

Mainframe