Paul Gilmartin wrote on 7/21/2024 10:07 AM:
But these things happen. I heard of a product that crashed reproducibly at customer sites having 8 or more tape drives.
Who does that in a test lab!>

This comes down to a problem with making sufficient resources available to developers and testers.  My sandbox system is a 4-way sysplex under VM.  There was an issue which looked like a bug but turned out to be an early design shortcoming from prior to my taking over that part of the product.  Some customers hit it on large systems but no matter how much I ran the tests on my sandbox, I could not trigger the problem.  Finally I got permission to try it on some other product's 7-way real LPAR system, and the problem hit on my first try.  And second try.  And third try.  So I was finally able to figure it out and fix the Day-0 design issue once I had sufficient resources available to encounter the problem.

Back when the model 3 IBM 3800 paper eater (*), er, high-speed laser printer, came out (1983 give or take), we got one, did the IOCP update, connected it up, fired it up and JES2 abended S0C4. Reliably if I recall.  I ended up in direct communication with the IBM developer who wrote the code.  He sent me an APAR fix and said "try this and let me know if works."  Me: "Didn't you test it?" Him: "No, I don't have a 3800-3 on my VM system."  The bug turned out to be that he didn't take into account that the -3 had longer sense data than the -1, so when JES2 did a SENSE to the -3, it overlaid a few bytes of some other critical JES2 data.  If the developer had had an actual 3800-3, he would have tripped over this very early on.

I rest my case.

(*) The 3800 moved the continuous form paper at 8 feet per second. Astounding opportunity for paper cuts if you ignored the warning to not run it with the cover open.

None of the above in any way justifies Clownstrike's shoddy code getting distributed.  I hope they get sued out of business.

/Leonard


----------------------------------------------------------------------
For IBM-MAIN subscribe / signoff / archive access instructions,
send email to [email protected] with the message: INFO IBM-MAIN

Reply via email to