I'd argue that the way businesses implement "agile" is in fact not agile. Just take SCRUM and/or SAFE and check what it imposes on you against the agile manifesto. They just don't do what they tell on their cover. Sprints are bad because they encourage devs to push out features that are not finished (if the sprint-end is approaching, testing is not done as thoroughly as may be required) and if it's discover during a sprint that a feature can be better implemented in a different way, that is not done as it would risk not finishing. Also, since Sprint-Goals are customer-value-driven, software developed in an "agile" manner does not receive a lot of cleanup and refactoring (as this does not directly yield "customer value"), which in the long run leads to heavily bloated Software.

what's the problem if get the job even by memorizing ?!

The stories I'm reading here sounds like an aberration of an idea of agile more than like agile. I'm a BA in agile team and to be honest im here to be on most of those meetings with businesses not developers. We have daili evey day and approximately 2-4 hours of refinements WEEKLY not daili. And since we decided that we are kind of mature team we have retro every second sprint or if needed. That said I think agile is fine as a ways of working but if you don't spend some time figuring out what you actually building and documenting that waterfall style you will have a bad time figuring things out before every sprint.

I’m scrum master and I also hate the current Agile implementation

Interesting take. Python replacing Cobol, Common Business Oriented Language, was designed to do fast and efficient transaction programming. Where the data format was set, say with an account number, name balance, type, interest rate,.... and you have a large finacial institution that has a tens of thousand customers and they need to process through those records nightly and update them. Back then it was mag tapes and you merged data from a data tape and an update tape onto an additional tape. Cobol is designed for that model of computing at its base. Add some sorting and then later when you had disks then data bases, you could replace the tape drives but not the transaction model. It is designed to do that job on large amounts of data with as little time as possible. A modern language like Python, or Java, or C# spends too much time frigging around with objects and data allocation and deallocation and such, that even given faster computers, if you match up Cobol and say Python, I think Cobol will win doing the large scale transactions day after day. Another language that had a special computational model is APL , A Programming Language, which was designed originally as a computer hardware definition language but because that model was based on matrix algebra, it turned out to be great for expressing mathematical matrix problems, like a log-linear regression in one line. But that language is do hard to understand it had the reputation of throw away code. It was easier and safer to just start over if you wanted to change a problem. Fortran is another example of things done right. It came out of the 50s but has been improved and was tuned to allow optimizing compilers to optimize the code strongly. So in a mathematical program Fortran now often runs 10 times faster than say an equivalent program in C++. So the issue really is, for each language there is a computational model it supports, a model of organizing a problem. Some are designed to be blindingly efficient, some are designed to take over all the storage allocations behind the scenes, some are designed to have a simple high level language that disguises whats under the cover. There is no one, perfect language and Computer Scientists, and programmers should have a deep view of the strengths and weaknesses of a language to understand where efficiencies and inefficiencies are. On another Historical note. In the 50s and early 60s, the first Databases came out. They were expensive, very expensive, but were adopted right away. Why? Because they broke the bond between a data field and its storage. You could define a record of fields like in a Cobol program and give it a size and type and the Database would take care of the storage and type and such. Previously, if you had say a Bank Account record for 100000 customers and you had the data record defined in each of say 1000 seperate Cobol programs, and someone came in with an larger last name, either from another country or Greek, or hyphenated... you had to change 1000 programs, and it might be that you only had the compiled program and not the source for some of them. Given a Database, and each program talking to the same record in the Database, you just have to change the Database definiton for that record and all the programs will work (except those that print it out in a limited space). But the key here, is why these languages exist, what are their strength, where are they appropriate. Fortram still out performs in its niche, Cobol outperforms in it's niche. Newer isn't necessarily better, but it is important that languages that are useful have programmers trained to use them.

Gooooood video

I may be 20 years out of date but I still know it. Also did some programming in IBM's Assembler Language. Probably have to relearn some of it for Z/OS due to new stuff being added.

3:28 I like your dog Dee.

For the extra GUI stuff I have but one comment: *Subroutines in more appropriate languages can be compiled with appropriate compilers (usually written in Fortran or COBOL) and then accessed from the Mainline code of the program*

I love COBOL. When I was taking programming me and my friend's husband were the only two who kept signing up for "COBOL on the World Wide Web", so they kept cancelling it.😢

This video brought back some memories! Back in 1982, I was US Army meteorologist stationed at Ft Greely, Alaska. The base was starting to become computerized and a number of units had been computerized by placing an IBM PC in a secure room and allowing it to connect to the mainframe IBM at headquarters. When I got to the base, I was put in charge of weather data entry from a couple dozen sites around the base. I would also generate reports at the end of the month for any officer who needed the data for their cold weather testing. There were a number of COBOL programs related to the data entry. There was one program that allowed the person entering data to enter a three digit site number, two digit month, and two digit year which got concatenated together into a seven digit file name and stored in a file on the disk. This program was 2000 lines long and took between 30 seconds and 2 minutes to execute,, depending on system load. The next COBOL program was the actual data entry and this was over 100,000 lines long and was incredibly slow. One set of weather data of around 8 parameters, 24 sets per day, took the same amount of time to process and save to disk as the site number. 24 sites X 24 sets of data per day X 30/31 days took a lot of time to enter. First day of the month I generated about 50 copies of everything on 11x14 fanfold paper, and there was a COBOL program for that, too. Start the process first thing in the morning and wait 2 days for the phone call from HQ notifying me that the copies were ready. All 200 pounds of them. I started by having the IBM PC take over the mainframe's duties. First I rewrote the program that made the file name into Batch language. The program was less than 20 lines and ran instantly. The first time the secretary ran the program (I had neglected to tell her that it was a bit faster), she picked up her coffee with her left hand as she hit the return key with her right hand and got the acknowledging (loud) beep instantly. She was so startled that she spilled half her coffee down her front! The next change was to convert the data entry part of the program. Batch was to kludgy for this so I converted it to Basic which would run locally on the PC and not on the mainframe. I used various musical notes from the PC's speaker to give acknowledgement of various conditions, such as ready, waiting, data stored, etc. The filename program got converted to Basic as well and added to it. Finally I converted the printout program and added it to the Basic program I was writing. Using some inline IBM Assembler code, I was able to print some snazzy graphics such as a large outline of Alaska on the title page. I would print out only one copy of everything to be used as a master and then I'd take the lot to the base copier to be duplicated into however many copies were needed. The copy machine would take only an hour to duplex print, collate, and staple the pages together. A rather amazing feat for a copier back in 1982. When I was done, the entire month's data could be entered in under a day, and saved many megabytes of data and hours of CPU time on the mainframe every month. For this, I was awarded the Army Commendation Medal.

Pay for other programming languages is significantly higher than Cobol

'Physician' is ripe for AI ... they are just glorified decision trees after all ;)

COBOL and JCL on IBM MF back in the day. FORTRAN, PASCAL, ASSEMBLER... done that at Central Texas College in 90's.

My mother was a COBOL programmer at Pratt and Witney in Montreal in the early 1970s. They were still using punch cards to run the programs, too. Turns out used programming cards were great for little kids to draw on. :-)

Any process can work if you have right people otherwise it is doomed to failure. It is that simple.

How valuable are COBOL programmers. Ask those who were around during the Y2K period. There was a lot of money to be made.

I took COBOL in college, late 1980s early 1990s. I used it in my first job but not since. I can do it again if the pay is right.But why Dont you get AI to do it?

cobol 68 and 74 pgmr...

I had to take it for my computer technology BS 20 years ago.

It is a solved problem why update an already solved problem?

I wonder if AI could be taught to program COBOL or convert it to a more usable language?

COBOL was, indeed, designed to be easy to read by the neophyte. Having nouns and verbs was revolutionary. There the clarity stops. Writing a compiler for the original COBOL version was very nearly impossible because the language is ambiguous: you can't really tell what a series of code paragraphs does simply by looking at it. A lexical scoping of the entire program must be done to determine how that paragraph is referred to is essential because of the dreaded "PERFORM abc THROUGH def" clause. This allows any arbitrary sequence of paragraphs to be called as a subroutine even if looking at the paragraphs in isolation gives no indication. This is just one factor that makes COBOL mathematically complex from a semantic viewpoint. There are lots of others. This is why the original COBOL code gets legacy status: the language is too complicated to write a reliable COBOL to language XYZ translator. Turns out it is much simpler and easier to write an IBM-360 machine language interpreter in some other language convenient on the destination environment. After all, computer vendors must have such a validation test for the architecture before they have the confidence to release a machine. The result is that the archaic COBOL source code lingers on. Like the cargo-cult worshipers did, we keep copies of the original lines of code and agonize over any proposed change. Prior to the IBM-360 series was the IBM-1401 series. It did have a FORTRAN compiler, not sure it had a COBOL translator. But the IBM-1401 did have machine language coding in the form of AUTOCODER. To this day, the binary AUTOCODER object code runs unchanged because the IBM-1401 architecture was also easy to emulate. Instead of making any changes to the AUTOCODER source, many installations just manually edit the object code with a binary text editor; that's really programming down in the dirt. I've written COBOL application suites and fiddled with some of the compiler implementations. What I like about COBOL is that it fills in a language I needed for the "C" spot in my resume language alphabet for a long time (A is for ASSEMBLY, ALGOL, and AUTOCODER, B is for BASIC, ..., Z is for Z-80 assembler; you get the idea.) Hope I haven't been obscure, really hope I haven't been obtuse, and don't mind if I get obliterated. Looking forward to any replies. Code in peace!

I missed an opportunity, back in 97/98 a MAJOR local banking company (M&I) was working with MSOE (or was it UWM?) to train cobol programmers. The course work was equivalent to masters degree and paid for by the bank (living expenses were your own responsibility, I had a wife and two children plus a mortgage) 25 years I regret not taking the opportunity. This was all in prep for Y2K.

Never knew Meghan Markle could code😮

I programmed in COBOL at university and in my first job;)

This is the perfect task for AI … imo . Have AI rewrite it to today’s language .

I coded in COBOL for 11 years. If coded well, COBOL could be written to be very clear and almost self-documenting. I then moved on to the AS/400 (IBM iSeries) and spent the rest of my career coding in RPGIII, RPGIV, and RPGILE. It was better suited to the IBM i. In college, I learned Pascal, Fortran and a whole host of weird languages. I even learned and did some coding (at home) in 6502 Assembly language. That was wild. I'm retired now but I get a ton of emails from companies looking for RPG programmers. The IBM i lives on.

what is this a HISTORY Lesson on COBOL .

Nice trip down memory lane, Thank you!

Great video!

In the 1990s when C++ was the new thing, we joked that there would be an object-oriented version of COBOL and it was going to be named “ADD 1 TO COBOL.”

PL/1 is my favorite, but I have coded and maintained some Cobol, but more Cobol 2. Old Cobol is extremely fragile. Any data problem ends with an abend. So if you have old Cobol use a newer and better compiler, and fix the errors. BTW Cobol II or later can interface with DB2, and CICS, etc. PL/1 allows data errors to be trapped and fixed on the fly. PL/1 used to run faster because its routines library is integrated into the operating system, so the executing code is much closer to machine level and does not have to run through the 370 machine code emulator to actually do work. More recently IBM may have fixed all of their mainframe languages to operate the same. All of the popular languages are interpretive, which makes coding and development easier, but all of them are too slow for processing billions of transactions. A language that is able to generate code which utilizes 32, 64, or 128 bit registers is needed for computations. Mainframe languages such as Assembler, Algol, Cobol, Pl/1, and maybe Ada (on a good day) could do most of that.

I learned COBOL 28 years ago and we made many financial and commercial programs, but I never used it professionnaly. If someone needs a COBOL programmer, I am ready, just reply to my comment. I'll invoice only $1,00.00 per code line.

In many cases, there just isn't a good reason to go through the risk and expense of migrating all of that code over to another language. I started programming in COBOL in 1969, and spent several years teaching it in a commercial environment. It's really not that hard to learn; like anything else, you just need to learn the basics and then see examples of practical applications. If you want to see something hard to learn, try Assembler. Tough to learn, but it was fast and it didn't take up a lot of memory, which was important back in the day. And so, people wrote really elegant code that wasted no resources. I've been retired for a few years now, and haven't actually coded in about 25 years, but I still remember the basics of COBOL.

I have a friend of mine that is still a highly proficient (but semi retired) programmer in COBOL and like me now in her 60's. She still comes out to play when her former customers need COBOL changes and makes a lot of money. When the Y2K panic came along in 2000 she was in big demand. Most old COBOL based software systems were never designed to go through a millennium change (00)She made a fortune upgrading COBOL based systems for many big companies, banks etc, and rich very quickly. She was quite busy for a year or two!! It is not really any harder to learn than any other language.

I bet the 40% of healthcare transactions that aren't COBOL are Mumps, which is somehow worse.

Cobol, Fortran, Assembler, remember them all

I learned cobol in the early 80s. It is not really that difficult.

I do and I disliked Cobol - I never put on my resume. I was offered a job with great pay and benefits and I had worked there helping facilitate the transfer between platforms. There was a system programmer's job but that job disappeared before my contract was. If I was hiring I would have taken that guy over me. When my contract was up I walked, I felt a bit sad because I really like their business social environment and I got to know a lot of the people on staff as I put on a number of mini-courses to teach the employees how to use a bunch of powerful tools now available to them to make their jobs easier, but there was no way I was going to stick around there and program in Cobol. That was in 1979.

The status if the Agile is failing or not depends on from whose perspective you are looking from. From devs perspective maybe yes, but then why dev should care about lost productivity/efficiency because of the extra meetings? Consider a meeting as a break and let managers deal with it. It is really not up to the dev to control this. From manager/customer perspective those extra meetings provide value, which is important to keep the project on track. Managers/customer get to know what is happening and can steer the project to correct direction early if needed. I really do not see why dev should be frustrated or feel bad about it, because it is part of the job. If there is less time for coding, then so be it. As a dev one also has to do demos, presentations, write solution designs, and do many other things.

I still can program in Cobol, even i know how to deal with Cobol and connect with databases like Oracle, MySQL, i don't see the problem here., The team lead is the one who needs to hire programmers with Cobol knowledge.

I learned COBOL, Common Business Oriented Language in 1975.

Hmmm. It seems the Computer Science aspects could learn a thing or two from the Soft Sciences. If they could get enough funding for proper Theory development and Replication, that is. "The money" has been tied to Engineering, without considering how the most fundamental problems (as with many processes in STEM) are sourced from poorly understood humanity. Consider "Publish or Perish" as an example.

If you can program COBOL FIS is offering a signing bonus. It’s not as dead as the media would have you believe.

2024 and we are still arguing over programming languages. Ridiculous! It's not about the language it's about the compiler. The world could use one language. Every programmer in the world could use the same language. Then choose the compiler based on the platform. Change my mind.

I am also a Chemical Engineer, and like you, I learned and used Fortran. On my first job, I wrote a program in Fortran to track the invoices and payments for a government contract that my company had. Our office was remote (180 miles) from the company HQ, thus it would have been more time consuming to use the computer team in the HQ. It probably should have been written in COBOL, however I was at the site and I knew Fortran, so fortran, it was.

I must have written over 200k lines of COBOL. I also used a COBOL code generator, where the best way to optimise it was by being proficient in Assember Macros. I do not miss any of it.

Not as old as Joe Biden.