Someone Asks me : "Will programmers be needed in the future if AI is already created code?"

Your question clearly tells one thing; you have no idea what programing is about and how it is done.

For starters, software already writes code. In every major codebase there are lots of files with auto-generated code. Auto-generated code is something a program have written to provide interface to a service for the rest of the codebase. So code already writes code - if the purpose of the code is clearly defined.

The problem is that users rarely know what they actually want, and even if they know it, they are rarely capable of creating a clear enough specification of what they want.

One step through the door my wife whips around, a look so disgusted she barely seems human. "What's that smell?" she cries. "It's you! You smell like...like bad code!"

Indeed, I am covered with the scent of the forbidden love child of a man who read half a chapter on if-then statements and then pushed out into the world, earthworm-like, a mangled misshapened gelatinous mass that my employer gave the title of line-of-business application purely out of pity.

For more days than I'd like to count I have been porting a ColdFusion 5 application to .NET. Initially written in 2000 and last touched in 2006, it has a data architecture comparable to Dresden after the second world war. It features a table solely comprised of seven columns of IDs so that joins can be made between other tables lacking a common key. Columns that should be contained within a single table spread out among multiple tables. Single columns containing data that should be multiple columns (with handy flags to separate the subsets). A view with 14 joins that playfully displays unintended results. And so much more spread out over almost 200 stored procedures, views, triggers, and tables on the SQL server, and dozens of additional ADO-like SQL statements within the ColdFusion itself. Fortunately, the application overcomes these issues by having absolutely no data validation while allowing nulls pretty much everywhere.

When I am done this will be a very nice ASP.NET MVC app with at least 150 less stored procs, views, and tables. Auto-generated duplicate entries will be a thing of the past. Pop-up windows that inexplicably refresh the underlying screen to display a different part of the program than the one the user wants will be eliminated. And a UI based on the colors of a Rubik's Cube with usability that Mr. Rubik would find challenging will disappear with only the trauma of using it left behind.

Sadly, this is not my worse legacy code experience. Just the most recent. Just the most recent stench added to a lifetime of bathing in code rot.

In our morning stand up, dev was bragging about how much code he was refactoring (like over-the-top bragging) and how much the changes will improve readability (WTF does that mean?), performance, blah blah blah. Boss was very impressed, I wasn't. This morning I looked at the change history and yes, he spent nearly two solid days changing code. What code? A service that is over 10 years old, hasn't been used in over 5, mostly auto-generated code (various data contracts from third party systems). He "re-wrote" the auto-generated code, "fixed" various IDisposable implementations and other complete wastes of time. How –bleep-ing needy are people for praise and how –bleep-ing stupid are people for believing such bull-bleep? I think I should get a t-shirt made with a picture of a BS-Meter and when he starts talking, “Wait a sec, I gotta change my shirt. OK…you were saying?”

Before becoming a developer, I used to work as a sales rep at this company that spent a good amount of time building what they believed to be an innovative state-of-the-art “code generator”. It was basically a scaffolding tool for generating software.

They were using it to auto generate customized iOS and Android native mobile app templates, along with a web backed.

The problem was that the generated code was shit, and the developers on the team basically spent more time fixing bugs than if they had built everything from scratch. But their passion for the product meant they just kept using it.

For some reason they never fixed issues in the original templates, so basically all the bugs that were found, kept showing up with each new app!

I have never seen apps like this that essentially had more bugs than features. Opening more than 10 app screen meant the app would freeze and crash. Sign up forms were actually dummy forms. The list goes on...

All the apps had the same shitty UI. For example, Product pages had a product image area that was like 5% of the screen view!

Last but not least, apps had a backend IP address hardcoded pointing to a server with an IP address that was temporary. So one day they had to restart the server and suddenly all customer apps stopped working and required a software update to work!

It was amazing seeing how a team of 3 developers trying to fix messy autogenerated code, couldn’t accomplish what was essentially a website on an app that I managed to build in my free time.

That’s how I knew it was time to quit my job and code full time.

WTF is the point a
of auto-generated documentation. Some dude litterally thought it was a good idea to read the code and write the exact same shit differently. WTF IS THE POINT!?

Documentation takes work, sorry, stop being lazy.

Pissed off. Planning on imposing a company wide hook that prevents you checking in code with a @Generated annotation. Seriously, never even heard about it being used outside of auto generated code until some bozos here seem to have started using it to silently drop complicated classes from test coverage metrics. Is this a thing with new coders these days, or are my lot just cowboys?!

No more, anyway. Sometimes it's convenient to be able to pull rank.

One thing I hate about WinForms:

"do not modify the contents of this method with the code editor."

How the phook do I add eventhandlers without adding them in the code? By some mysterious ways in the so called "designer"? Click and drag? Wearing a tie and talking to the computer? Making a phone call to Microsoft? Immersing into VR and ordering an event handler from a virtual store?

No, I've always just typed whatever I want to achieve and I'm going to do so whatever those know-it-all comments say in the auto-generated code.

You can call me a conservative old fart, but nothing beats writing my own code.

Worst architecture I've seen?

The worst (working here) follow the academic pattern of trying to be perfect when the only measure of 'perfect' should be the user saying "Thank you" or one that no one knows about (the 'it just works' architectural pattern).

A senior developer with a masters degree in software engineering developed a class/object architecture for representing an Invoice in our system. Took almost 3 months to come up with ..

- Contained over 50 interfaces (IInvoice, IOrder, IProduct, etc. mostly just data bags)
- Abstract classes that implemented the interfaces
- Concrete classes that injected behavior via the abstract classes (constructors, Copy methods, converter functions, etc)
- Various data access (SQL server/WCF services) factories

During code reviews I kept saying this design was too complex and too brittle for the changes everyone knew were coming. The web team that would ultimately be using the framework had, at best, vague requirements. Because he had a masters degree, he knew best.

He was proud of nearly perfect academic design (almost 100% test code coverage, very nice class diagrams, lines and boxes, auto-generated documentation, etc), until the DBAs changed table relationships (1:1 turned into 1:M and M:M), field names, etc, and users changed business requirements (ex. concept of an invoice fee changed the total amount due calculation, which broke nearly everything).

That change caused a ripple affect that resulted in a major delay in the web site feature release.

By the time the developer fixed all the issues, the web team wrote their framework and hit the database directly (Dapper+simple DTOs) and his library was never used.

Over the summer I was recruited to be a supplement instructor for a data structures course. As a result of that I was asked (separately by the professor) to be a grader for the course. Because of pay limitations I've mostly been grading homework project assignments. In any case, it's a great job to get my foot into the department and get recognized.

Over the course of the semester I've had this one person, OSX, named after their operating system of choice, who has been giving me awkward submissions. On the first assignment they asked the professor for extra time for some reason or the other, and that's perfectly fine.

So I finally receive OSX's submission, and it's a .py file as per course of the course. So I pop up a terminal in the working directory and type "python OSX_hw1.py". Get some error spit out about the file not being the right encoding. I know that I can tell python to read it in a different encoding, so I open it up in a text editor. To my surprise it's totally not a text file, but rather a .zip file!

I've seen weirder things done before, so no big deal. I rename the file extension, and open it up to extract the files when I see that there's no python files. "Okay, what's goin on here OSX..." I think to myself.

Poking around in the files it appears to be some sort of meta-data. To what, I had no clue, but what I did find was picture files containing what appeared to be some auto-generated screenshots of incomplete code. Since I'm one to give people the benefit of doubt even when they've long exhausted other peoples', I thought that it must be some fluke, and emailed OSX along with the professor detailing my issue.

I got back a rather standard reply, one of which was so un-notable I could not remember it if my life depended on it. However, that also meant I didn't have to worry about that anymore. Which when you're juggling 50 bazillion things is quite a relief. Tragically, this relief was short lived with the introduction of assignment 2.

Assignment 2 comes around, and I get the same type of submission from OSX. At this time I also notice that all their submissions are *very* close to the due time of 11:59pm (which I don't care about as long as it's in before people start waking up the next morning). I email OSX and the professor again, and receive a similar response. I also get an email from OSX worried about points being deducted. I reply, "No issue. You know what's wrong. Go and submit the right file on $CentralGradingCenter. Just submit over your old assignment".

To my frustration OSX claimed to not know how to do this. I write up a quick response explaining the process, and email it. In response OSX then asks if I can show them if they comes to my supplemental lesson. I tell OSX that if they are the only person, sure, otherwise no because it would not be a fair use of time to the other students.

OSX ends up showing up before anyone else, so I guide them through the process. It's pretty easy, so I'm surprised that they were having issues. Another person then shows up, so I go through relevant material and ask them if they have any questions about recent material in class. That said, afterwards OSX was being somewhat awkward and pushy trying to shake my hand a lot to the point of making me uncomfortable and telling them that there's no reason to be so formal.

Despite that chat, I still did not see a resubmission of either of those two assignments, and assignment 3 began to show it's head. Obviously, this time, as one might expect after all those conversations, I get another broken submission in the same format. Finally pissed off, I document exactly how everything looks on my end, how the file fails to run, how it's actually a zip file, etc, all with screenshots. That then gets emailed to the professor and OSX.

In response, I get an email from OSX panicking asking me how to submit it right, etc, etc. However, they also removed the professor from the CC field. In response I state that I do not know how to use whatever editor they are using, and that they should refer to the documentation in order to get a proper runnable file. I also re-CC the professor, making sure OSX's email to me is included in my reply.

OSX then shows up for one of my lessons, and since no one had shown up yet, I reiterate through what I had sent in the email. OSX's response was astonished that they could ever screw up that bad, but also admits that they had yet to install python(!!!). Obviously, the next thing that comes from my mouth is asking OSX how they write their code. Their response was that they use a website that lets them run python code.

At this point I'm honestly baffled and explain that a lot of websites like those can have limitations which might make code run differently then it should (maybe it's a simple interpreter written on JavaScript, or maybe it is real python, but how are you supposed to do file I/O?) .

After that I finally get a submission for assignment 1!

I was just wondering why the flying fuck OxyPlot was rendering an unknown node on my chart for monday when I only queued my sql db for a single Tuesday.

Noticed it was an auto generated node, which had the right time, status, everything.

I looked up my code and found out that the date somehow bugged out, and if c# isn't sure what to do with an empty date it initializes a date on 1.1.0001

And what do you know. Appearantly thats a Monday.

I broke my head way harder over this than I should have, assuming the worst possible bugs in the render engine. fml

There was a big hairy ball of SW mud from another project that a poor coworker had to "reuse". Only that it was impossible because there was no documentation, shit was partly auto-generated with mysterious Excel tables, and the actual code was just as bad. No APIs and nothing, just hacking shit into globals, several nested state machines that were overriding each other's states, and with global side effects. WTF.

Two devs took a look at it - minimum 8 weeks. Schedule was some days, and PM insisted that it was "already working". But the worst thing was that the dev in charge had been looking for another job anyway and quit, so the whole clusterfuck suddenly was on my desk.

The code was so awful that I could only bear it with both eyes closed, so I instead read the spec of this project closely. Turned out that it didn't actually demand this feature, only a small subset of what the ball of mud was supposed to achieve - which I was able to implement from scratch within a day, plus another one for documentation. Phew.

a previous employee used a CMS called pimcore to develop an internal application, using only the admin interface.
now my supervisor wants me to implement a custom functionality that the system doesn't quite offer(querying an external database) in a way so that it doesn't interfere with the rest od the app; so I have to filter through a clusterfuck of auto-generated code and UI to find where, how and why components of the page are generated.
don't get me wrong, pimcore seems to be a solid CMS, but it's just the wrong tool for the job.

Really fucking nice android studio. Your auto generated code makes my builds fail and your error message don't even care enough to show me in which direction I should go fuck myself.

1) Had to fix severe bugs in a dynamic UI (configuration-driven forms) component.

Recognized undocumented Copy/Paste/Modify/FuckUp driven variations of the same component all over the project. Unsurprisingly, the implementations covered 99% of the antipattern catalog on wiki.c2.com and could compete with brainfuck in regard to human-readable code.

Escalated the issue, proposed a redesign using a new approach, got it approved.

Designed, Implemented, tested and verified the new shared and generic component. Integrated into the main product in the experimental branch. Presented to tech lead/management. Everyone was happy and my solution opened even more possibilities.

Now the WTF moment: the product with the updated dynamic UI solution never has been completely tested by a QA engineer despite my multiple requests and reminders.

It never got merged into baseline.
New initiatives to fix the dynamic UI issues have been made by other developers. Basically looking up my implementation. Removing parts they do not understand and wondering why the data validation does not work. And of course taking the credit.

2) back in 2013, boss wanted me to optimize batch processing performance in the product I developed. Profiling proved that the bottleneck ist not my code, but the "core" I had to use and which I must never ever touch. Reported back to him. He said he does not care and the processing has to get faster. And I must not touch the "core".

(FYI: the "core" was auto-generated from VB6 to VB.Net. Stored in SourceSafe. Unmaintainable, distributed about a bunch of 5000+ LoC files, eye-cancer inducing singlethreaded something, which had naive raw database queries causing the low performance.)

personal projects, of course, but let's count the only one that could actually be considered finished and released.

which was a local social network site. i was making and running it for about three years as a replacement for a site that its original admin took down without warning because he got fed up with the community. i loved the community and missed it, so that was my motivation to learn web stack (html, css, php, mysql, js).

first version was done and up in a week, single flat php file, no oop, just ifs. was about 5k lines long and was missing 90% of features, but i got it out and by word of mouth/mail is started gathering the community back.

right as i put it up, i learned about include directive, so i started re-coding it from scratch, and "this time properly", separated into one file per page.

that took about a month, got to about 10k lines of code, with about 30% of planned functionality.

i put it up, and then i learned that php can do objects, so i started another rewrite from scratch. two or three months later, about 15k lines of code, and 60% of the intended functionality.

i put it up, and learned about ajax (which was a pretty new thing since this was 2006), so i started another rewrite, this time not completely from scratch i think.

three months later, final length about 30k lines of code, and 120% of originally intended functionality (since i got some new features ideas along the way).

put it up, was very happy with it, and since i gathered quite a lot of user-generated data already through all of that time, i started seeing patterns, and started to think about some crazy stuff like auto-tagging posts based on their content (tags like positive, negative, angry, sad, family issues, health issues, etc), rewarding users based on auto-detection whether their comments stirred more (and good) discussion, or stifled it, tracking user's mental health and life situation (scale of great to horrible, something like that) based on the analysis of the texts of their posts...

... never got around to that though, missed two months hosting payments and in that time the admin of the original site put it back up, so i just told people to move back there.

awesome experience, though. worth every second.

to this day probably the project i'm most proud of (which is sad, i suppose) - the final version had its own builtin forum section with proper topics, reply threads, wysiwyg post editor, personal diaries where people could set per-post visibility (everyone, only logged in users, only my friends), mental health questionnaires that tracked user's results in time and showed them in a cool flash charts, questionnaire editor where users could make their own tests/quizzes, article section, like/dislike voting on everything, page-global ajax chat of all users that would stay open in bottom right corner, hangouts-style, private messages, even a "pointer" system where sending special commands to the chat aimed at a specific user would cause page elements to highlight on their client, meaning if someone asked "how do i do this thing on the page?", i could send that command and the button to the subpage would get highlighted, after they clicked it and the subpage loaded, the next step in the process would get highlighted, with a custom explanation text, etc...

dammit, now i got seriously nostalgic. it was an awesome piece of work, if i may say so. and i wasn't the only one thinking that, since showing the page off landed me my first two or three programming jobs, right out of highschool. 10 minutes of smalltalk, then they asked about my knowledge, i whipped up that site and gave a short walkthrough talking a bit about how the most interesting pieces were implemented, done, hired XD

those were good times, when I still felt like the programmer whiz kid =D

as i said, worth every second, every drop of sweat, every torn hair, several times over, even though "actual net financial profit" was around minus two hundred euro paid for those two or three years of hosting.

I need to compare the JSON results of an API before and after a code change. But it was also moved to another API.

However some fields are auto-generated like timestamp or derived off the url (resource links).

Also if a JSON list is returned it maybe in different order...

Wondering is there a quick way to test text likeness?

I've done it before but just used matching status code and maybe measuring the diff in response size

rent / question (there is a question at the end and I'd appreciate your opinion)

8 months ago, I agreed to help a not too distant relative of mine to do his master thesis at the company where I work. He was supposed to build something really MVP, but useful for us and I'd help him get some scientific questions out of it, and provide him with (computing) resources to test his theories / implementations under simulated and much heavier load.

Since then, he didn't get done anything even remotely useful, always just stuck on very rudimentary issues, claimed things are almost ready, I wrote a quick smoke test to prove that the whole application blows up when you touch it, in short - a disaster and went over to radio silence.

In the meanwhile, we didn't need it anymore, so 1.5 months ago, I got in touch with him again, with an even more technical proposal, something, at least I'd think, that's even cooler to do. He asked me some question about hypothetical load, the system should be able to handle eventually, to come up with alternative implementations to compare them against each other. He said that his exam period is going to be over soon and he'll get back to me with some initial version.

2 weeks ago, I got back in touch with him, trying to urge him, to get finally started and get something done. If he'd actually sit down and do it during the holidays as a "full time job", he'd be probably done in 2 weeks. Last week, he came back to me and said he has an initial PR ready to review.

I was excited about it, but basically froze when I realized what he did. He deleted all his previous work - some infrastructure stuff which took us basically 3 months of back and forth to get running - and as far as I could see, all the new code were only auto generated clients based on a swagger specification. In short - I could do it in less then an hour. If you really have no idea what you're doing, it might take you half a day, but definitely nowhere near to a week.

His brother, which a good friend of mine, thinks I'm being too hard on him. His argument was, that it's too hard, and he has to do it in C#, but he only knows Java (I gave him access to some of our repositories to copy paste code together, he didn't need to invent anything. I also prefer C# but wrote my master thesis in Java) Personally, I'm just pissed because he promises stuff that he never does. I totally understand him - I was like that as a student as well, I guess karma is a ... but still, he's wasting my time.

Right now I'm thinking how to get out of this, without having even more time wasted. I doubt he'd ever deliver anything useful. He got plenty of input from me about what he could consider for his scientific question, how to measure performance, ... He can keep his credentials to access our test environment with the test data, but I won't give him access to any additional computing resources, to compare how his solutions might scale on our company's cost. (mainly it's not the money, but I'd have to provide that stuff, and probably help him set it up)

does it sound like a fair deal (saying, I'm done with you. You can finish your topic on your own, but don't expect any help from me)? or am I being a dick about it and too demanding?

Fuck undocumented shit!
I was wondering how to use this one method of the "interface". Googled the name of the program and the specific method.

Got two results.

From the same page.

It's a comment complaining that this shit is undocumented and doesn't work.

If you build something that others use, please, motherfucking please, document your code.
At least some auto generated javadoc, how hard can it be?
You are using the atlassian suite for everything and you have confluence so use it already! The only documentation that actually exists it about a hundred years old, totally useless and covers about 1% of what your product can do.
I like your product but fuck me sideways your documentation sucks balls! Fuck!

That needed to get out.

I need help structuring a new TypeScript project built on a MERN stack. I used CRA for the client, so I opted to have separate tsconfig files -- one for client (auto-generated by CRA) and one for server (extends node12 tsconfig). However, I'm trying to setup eslint and prettier globally so that the lint/style rules are uniform across the codebase. CRA adds an eslint config that extends react-app, which is fine, but I'd like to still have my global rules. I have written my eslintrc.json file and am happy with it, so I placed it in the project root directory. I figured I would install eslint, prettier, etc. in the project root, then when I run eslint globally, it would lint the server code with the global rules and the client code with the global rules and the react-scripts rules.

However, react-scripts complains that I've installed a newer version of eslint in a parent directory. I can either ignore that rule or use the same version as react-scripts, but it seems like react-scripts is going to run eslint on its own when I run npm start, regardless of if I have a global config. What should I do? Is there a better way to structure the app?