How do we improve dump analysis?

Hello Fellow Forum Members,

I need your help!

The Broadcom incubator team is researching ways to improve Mainframe dump analysis. We’re considering how to accelerate troubleshooting, increase the productivity of junior developers, and decrease the demand on senior developers.

Are you willing to tell us your challenges? If yes, learn more and complete a short survey.

Thanks for your time!


hI Praveen
Lets have more detailed and on line meetings over this to give a joint review of the problem statement.

  1. On the fly, suggest a few means to make it more simpler

  2. Write a rex which would take sysabend filea s an input and pronounce the errors in a much user friendly manner so that the coders wont run away from looking at the dunp dataset

  3. Use assembler language to work like AI or machine language to prompt out the error codes with a brief explanation on what went wrong and what actions to take when a RC> 0 is recieved on the outlist or sdsf?

  4. Lets replace the jcl for compile/precompile/ bind/ link edit with icons being driven by menu driven panels or mouse enabled for background…The coders need to focus on the application logoc rather than the jcl stuffs.

1 Like

Hi Kapitan,

Thanks for your suggestions! If you have more time we would like to talk with you. Please let me know if you are interested in a meeting. Please use this link to find a time and data that suits you the most -

Praveen Verma

Short answer … Don’t write code that fails … Seriously, it’s really an engineering challenge to software developers. There are Unix programs that never abend … So on Z/OS you have to take care to abide some reverent rules that seem to escape most newbies. I’m an Assembler programmer by trade who has maintained mainframe software for 35 years. To answer your obvious questions on how to write programs that don’t abend would take too much time, although it’s quite simple in theory, not to hard in practice.