When I were a lad...

[pauldee]

You would not have survived in core banking days or cobol and RPG - code still running 85% of all banks world wide


Sent from my iPhone using Contractor UK Forum

I expect I would have been fine. You're comparing things that happened decades ago with modern coding practices.

A line of code every 3 seconds is code churn however you look at it, and if you really believe that you couldn't have taken a minute or two to think things through and come up with possibly better alternatives then you are deluding yourself.

There may still be a lot of Cobol and RPG around but banks are slowly and painfully replacing it in order to stay competitive with the more agile Monzos and fintechs.

[jayn200]

That's one line of code every 3 seconds, non stop, for 8 hours. I can barely type that fast, so I struggle to think that you were stopping to think about whether you were applying the best approach.

When picking up an agile ticket I can spend a good hour exploring the various possible solutions, and not typing any code: which is most performant? Which will be the most maintainable? Which will work best with with future features coming up?

Then maybe refactoring existing code to prepare for the new code, then refactoring afterwards to keep the code base clean. Often I will end up with fewer lines of code than when I started, but with a new feature added.

You're thinking of 10,000 lines of code in a day, he did say in a week. So that's only 1 line of code per 13.5 seconds or so.

Still no one is writing 10,000 lines of original code in a week. You can easily generate 10,000 lines of code using some form of auto generation or copy/pasting. Considering GhostofTarbera said "back in the day" I am going to guess he was copy pasting mass amounts of code.

[pauldee]

You're thinking of 10,000 lines of code in a day, he did say in a week. So that's only 1 line of code per 13.5 seconds or so.

Still no one is writing 10,000 lines of original code in a week. You can easily generate 10,000 lines of code using some form of auto generation or copy/pasting. Considering GhostofTarbera said "back in the day" I am going to guess he was copy pasting mass amounts of code.

You're absolutely right, 10000 lines of code per week, back in the day.

He didn't seem seem to pick me up on the every 3 seconds though, and apparently the teams he PMs are turning out 10 lines of code per week. I'd probably suggest where to start looking for that issue as I'm not seeing it anywhere.

I did just assume he was exaggerating but then bedded in with the 'fully optimised' thing.

One line of code every 13.5 seconds still doesn't suggest much thoughtful consideration on the code produced.

[eek]

You're absolutely right, 10000 lines of code per week, back in the day.

He didn't seem seem to pick me up on the every 3 seconds though, and apparently the teams he PMs are turning out 10 lines of code per week. I'd probably suggest where to start looking for that issue as I'm not seeing it anywhere.

I did just assume he was exaggerating but then bedded in with the 'fully optimised' thing.

One line of code every 13.5 seconds still doesn't suggest much thoughtful consideration on the code produced.

I wonder if this helps at all

People Speak Faster in Less Efficient Languages - The Atlantic

Some languages are more long winded than others so it may take a lot longer to right something in one language rather than another.

Regular Expressions are a prime example - it could save you 1000 lines of code but it will take you the same length of time to ensure the regular expression is correct.

[pauldee]

I wonder if this helps at all

Daily chart - Why are some languages spoken faster than others? | Graphic detail | The Economist

Some languages are more long winded than others so it may take a lot longer to right something in one language rather than another.

Regular Expressions are a prime example - it could save you 1000 lines of code but it will take you the same length of time to ensure the regular expression is correct.

Fair point, functional languages especially seem to cram a lot on to one line, but takes some thinking about.

Like I said, number of lines of code is a bad metric.

If someone I worked for was daft enough to pay developers by number of lines of code, I would probably be churning out 10000 lines per day. It would be crap code though.

[PlanB]

Fair point, functional languages especially seem to cram a lot on to one line, but takes some thinking about.

Like I said, number of lines of code is a bad metric.

If someone I worked for was daft enough to pay developers by number of lines of code, I would probably be churning out 10000 lines per day. It would be crap code though.

Good point about functional languages, it might take me a good 10-15 mins pondering a line of Haskell, but it is packing in a load of work. If you drop down and start churning out ASM I can see how you would need hundreds of lines of code to achieve the same thing.

10K of good code in a week, week in and out is a physical impossibility, the coding world has moved on and you need to make code readable and maintainable as that is going to be the majority of it's life. Making it work is just the start of it, you need refactoring and accompanying unit tests.

[GhostofTarbera]

Good point about functional languages, it might take me a good 10-15 mins pondering a line of Haskell, but it is packing in a load of work. If you drop down and start churning out ASM I can see how you would need hundreds of lines of code to achieve the same thing.

10K of good code in a week, week in and out is a physical impossibility, the coding world has moved on and you need to make code readable and maintainable as that is going to be the majority of it's life. Making it work is just the start of it, you need refactoring and accompanying unit tests.

And back in the day we did not have

Product owners
Project managers
Test managers or testers
Scrum idiots
Infra guys
QA

We did all those jobs ourselves, And it all seemed to work, I can count on one hand bug fixes, now bug fixing is the norm for bad programming and testing


Sent from my iPhone using Contractor UK Forum

[pauldee]

Good point about functional languages, it might take me a good 10-15 mins pondering a line of Haskell, but it is packing in a load of work. If you drop down and start churning out ASM I can see how you would need hundreds of lines of code to achieve the same thing.

10K of good code in a week, week in and out is a physical impossibility, the coding world has moved on and you need to make code readable and maintainable as that is going to be the majority of it's life. Making it work is just the start of it, you need refactoring and accompanying unit tests.

Good point about tests, I doubt the original 10000 lines per week in Cobol included tests, but today do you count the tests code lines in the count? They're not actually doing anything functionally productive but they do save a lot of work in the long run. Most decent developers write considerably more lines of test code than production code.

[PlanB]

And back in the day we did not have

Product owners
Project managers
Test managers or testers
Scrum idiots
Infra guys
QA

We did all those jobs ourselves, And it all seemed to work, I can count on one hand bug fixes, now bug fixing is the norm for bad programming and testing


Sent from my iPhone using Contractor UK Forum

I started in 97 in much the same environment in a finance house, the code was a bug ridden mess hacked together by a room full of C and Cobol contractors on eye watering day rates. The only bugs recorded were the ones in prod where segfaults took the system down, there were many, many un-recored ones as there was zero testing done by the dev team.

One enterprising chap knocked up a shell script to identify the dangling pointers and memory leaks that riddled the code and came back with a list of about 50 for us to work through.

[eek]

I started in 97 in much the same environment in a finance house, the code was a bug ridden mess hacked together by a room full of C and Cobol contractors on eye watering day rates. The only bugs recorded were the ones in prod where segfaults took the system down, there were many, many un-recored ones as there was zero testing done by the dev team.

One enterprising chap knocked up a shell script to identify the dangling pointers and memory leaks that riddled the code and came back with a list of about 50 for us to work through.

All Tabera's post remind me of the Post Office's Horizon project - because the code worked on one level no-one cared about the fundamental faults that existed within it.

And all bank software was the same - code where all bugs except the incredibly obvious lay unknown waiting to spring out and catch people out..