I knew punch cards would return!

[Don Shepherd]

Ah, the IRS returning to its Univac days of yore.

With vinyl records making a comeback, can IBM punch cards be far behind?

Dust off that COBOL code.

IRS problem

[Paul Dale]

They've made a few comebacks already.


Pauli

[david sanz]

(04-19-2018 07:26 AM)Don Shepherd Wrote:  Ah, the IRS returning to its Univac days of yore.

With vinyl records making a comeback, can IBM punch cards be far behind?

Dust off that COBOL code.

IRS problem

It is my understanding that most banks are still using COBOL today, at least in Spain.

Seems that something similar is going on in IRS.

While we are talking about governments and computers, I can tell you that most Spanish governmental websites use Java based apps, which very often stop running after Java updates or are blocked by Java because of issues with security certificates.

[Paul Berger (Canada)]

(04-19-2018 12:11 PM)david sanz Wrote:  

(04-19-2018 07:26 AM)Don Shepherd Wrote:  Ah, the IRS returning to its Univac days of yore.

With vinyl records making a comeback, can IBM punch cards be far behind?

Dust off that COBOL code.

IRS problem

It is my understanding that most banks are still using COBOL today, at least in Spain.

Seems that something similar is going on in IRS.

While we are talking about governments and computers, I can tell you that most Spanish governmental websites use Java based apps, which very often stop running after Java updates or are blocked by Java because of issues with security certificates.

I believe this is pretty common in the financial sector and in some cases the problem boils down to the source for an application being lost, but often it is due to a "leave well enough alone mentality" and when changes are needed they look for solution that require the minimum amount of change. A friend who worked as a programmer for a large insurance company told me they resolved the Y2K problem in a large database on an IBM mainframe by a simple change. It had been designed with a two digit year which was stored in packed decimal which would require two bytes because of the sign nibble and one digit nibble would always be zero so for anything after 2000 they simply changed that nibble to a non-zero value which only require minimal changes to the program and none to the database itself.

[Thomas Okken]

(04-19-2018 02:38 PM)moonbeam Wrote:  

(04-19-2018 12:11 PM)david sanz Wrote:  It is my understanding that most banks are still using COBOL today, at least in Spain.

It wouldn't surprise me if there were more lines of COBOL code running on production systems today than those of all other programming languages combined.

Exactly.

I used to work for a large American phone company, and the middle tier systems ran Unix with application code (in K&R C) going back to the '70s, and the back-end systems were IBM mainframes running databases and business logic (COBOL) that were even older.

The general rule was: never rewrite anything. It's not that they're in the habit of losing their source code, it's because starting over would be insanely expensive, and debugging all those millions of lines of new code would be hellish-to-impossible.

They did do a pretty thorough job for Y2K compatibility. Changing 2-digit years to 4-digit, checking business logic everywhere, and running endless regression tests. When the new millennium finally arrived, nothing bad happened.

Youngsters often scoff at what they see as bloated or crusty bureaucracies, but by god, these guys knew how to keep things running. I have since worked for a company that has more of a start-up mentality, and it shows: we crank out new functionality faster, but we also have more downtime. Take your pick.

[Gene]

Did a lot of COBOL programming in DB2 and IMS DB/DC environments in the early 1990s. We had to design the systems for the state to be Y2K compliant. I'm sure the COBOL is still being run as object code of course.

MOVE YEAR-MONTH-CODE TO Y2K-YEAR-MONTH-CODE.

blah.

[Don Shepherd]

(04-19-2018 02:50 PM)Paul Berger (Canada) Wrote:  A friend who worked as a programmer for a large insurance company told me they resolved the Y2K problem in a large database on an IBM mainframe by a simple change. It had been designed with a two digit year which was stored in packed decimal which would require two bytes because of the sign nibble and one digit nibble would always be zero so for anything after 2000 they simply changed that nibble to a non-zero value which only require minimal changes to the program and none to the database itself.

While preparing for a talk on IBM punchcard systems for HHC2017 I discovered that the US Army, during WW2, used IBM tabulation and unit record systems for all kinds of things, and they frequently stored the month code in one column in the IBM card. And sometimes they stored the year code in one column as well (perhaps hoping or assuming that the war would be over before 1950 I suppose). 80-column IBM cards were great for their time, but when you only have 80 columns you have to compromise. And when magnetic disks came along, you still compromised because disk storage wasn't cheap.

I hope I am alive in 9999 to warn of the dangers inherent in the 4-digit year!

[DavidM]

(04-19-2018 03:23 PM)Thomas Okken Wrote:  Youngsters often scoff at what they see as bloated or crusty bureaucracies, but by god, these guys knew how to keep things running. I have since worked for a company that has more of a start-up mentality, and it shows: we crank out new functionality faster, but we also have more downtime. Take your pick.

Along these same lines, I believe we are being conditioned over time to accept downgraded service and performance in many types of technology areas. The increasing complexity of embedded systems (whether in calculators, cars, or cell phones) makes flawless operation an increasingly rare occurrence.

Most of us are old enough to remember times when cell phones didn't exist. If we had had as poor an experience with unclear audio, dropped calls, blended signals and the like with "regular" POTS service as we routinely do now with cell service, there would have been public outrage. Now it is simply accepted as the state of the art for the technology.

This is only part of why I am very skeptical about self-driving cars...

[Massimo Gnerucci]

Of course COBOL is here to stay.

Maybe I'll end my career where it all began...

[Dave Britten]

Good thing I just got MS COBOL up and running on my Altairduino!

Not that I know how to use it.

Yet.