I am fed up working with unskilled software developers. Or to be more specific, working with people who have no idea of sofware architecture.

Most people I've worked with have simply no idea what they are doing in the broad picture, they can only follow patterns they see and implement their feature in the same way. They can't think about the abstract concepts which should be the foundation of the project.

They fail to write unit tests which are maintainable. They write one fucking test per method which is testing 50 things at the same time, making it often impossible to understand what is being tested.

They think putting stuff in private methods makes their class better and is some kind of separation of concerns.

They write classes and afterwards create interfaces for these classes named {Class}Interface, shoving all the methods into that interface. They think it's good design to do so.

They are unable to think about the reasons why things are done the way they are done and that you don't do stuff for the sake of doing stuff, but to achieve certain goals like interchangeability.

They don't undestand how to separate business logic from the application code.

They have no sense for naming things beautifully. They don't see how naming things is a major part of good software architecture.

They get layer concepts wrong and then create godlike {EntityName}Service classes, which do everything related to a particular entity.

They fail to shape the boundaries within a software project, entangling stuff which should live in individual modules.

All I want is to work in a team with professionals.

Legacy code.

Honestly though, this is some of the better legacy code I've worked with at this company. It's a nifty alert system wherein you can trigger sending messages to subscribers of that alert via whatever means (phone/email) they've entered.

I'll save you the technical analysis of its internals, but suffice to say it's actually pretty nice, with good separation of concerns, internal logic hidden away, dead-simple public interface, etc. documentation is kinda crap, but it exists (!), so that's a nice change.

but.

For some unknown and bloody bizarre reason, the thing breaks when a user wants both sms AND email notifications. Either by themselves work totally fine, but both together? nonono. Email alerts give ArgumentErrors, so something internal isn't correct, and SMS alerts complain about uninitialized Twilio::Error constants.

but.
they both work fine otherwise?

also, the two notification preferences aren't stored on the same object anywhere. if a user wants both, the user creates two AlertContact objects with different info, and when performed, the Alert basically iterates over these and does its thing for each, so there is no knowledge shared between them. totally should work the same regardless.

idfgi.

ALSO.
AND THIS PART REALLY PISSES ME OFF.

WHEN THERE'S AN ERROR, THIS THING DOESN'T LOG IT. IT STRINGIFIES THE ERROR OBJECT (basically just extracting the message) AND INSERTS THAT INTO THE DATABASE INSTEAD. WHAT THE CRAP.

So, I don't get a stack trace, line number, or anything. just the basic error message. instead of my alert text. because of course that makes sense and totally helps debugging.

aklsjfak;sldfj.
legacy code.

Just need to get this off my chest. Started a new job 3 weeks ago at a company that has been around ~18 years, it is only recently that they have started to grow more rapidly. I was brought in under the guise that they wanted to embrace change and better practices and so said I was up for the challenge.

In my 2nd week I was asked to produce a document on tackling the technical debt and an approach to software development in the future for 3 consultants who were coming in to review the development practices of the company on behalf of the private equity firm who has taken a major stake in the company. I wrote the document trying to be factual about the current state and where I wanted to go, key points being:

Currently a tightly coupled monolith with little separation of concerns (73 projects in one solution but you have to build two other solutions to get it to build because there are direct references.).

Little to no adherence to SOLID principles.

No automated testing whatsoever.

Libraries all directly referenced using the file system rather than Nuget.

I set out a plan which said we needed to introduce TDD, breaking dependencies, splitting libraries into separate projects with nuget packages. Start adhering to SOLID principles, looking at breaking the project down into smaller services using the strangler pattern etc. After submitting what I had written to be part of a larger document I was told that it had been tweaked as they felt it was too negative. I asked to see the master document and it turns out they had completely excluded it.

I’ve had open and frank discussions with the dev team who to me have espoused that previously they have tried to do better, tackle technical debt etc but have struggled to get management to allow them. All in all a fairly poor culture. They seem almost resigned to their fate.

In my first 2 weeks I was told to get myself acquainted and to settle myself in. I started looking at the code and was quite shocked at how poorly written a lot of it was and in discussions with my manager have been critical of the code base and quite passionate and opinionated about the changes I want to see.

Then on Friday, the end of my third week, I was invited to a meeting for a catch up. The first thing I was told was that they felt I was being too openly critical in the office and whether I was a good fit for the company, essentially a stay or go ultimatum. I’ve asked for the weekend to think about it.

I’ve been a little rocked by it being so quickly asked if I was a good fit for the company and it got my back up. I told them that I was a good fit but for me to stay I want to see a commitment to changes, they told me that they had commitments to deliver new features and that we might be able to do it at some point in the future but for now I just needed to crack on.

Ordinarily I would just walk but I’ve recently started the process to adopt kids and changing jobs right now would blow that out the water. At the same time I’m passionate about what I do and having a high standards, I’m not going to be silenced for being critical but maybe I will try and tackle it in a different way. I think my biggest issue is that my boss who was previously a Senior Developer (my current position) has worked at the company for 12 years and it is his only job, so when I’m being critical it’s most likely criticising code he wrote. I find it hard to have the respect of a boss who I had to teach what a unit test was and how to write one. It makes it hard to preach good standards when by all accounts they don’t see the problems.

Just wondering if anyone has suggestions or experience that might help me tackle this situation?

Good documentation is always a fucking good experience man.

And I particularly like how the Vue.js documentation is laid out. Straight up the framework is:
Easy to use
Concise
Has a lot of sane ideas
Good separation of concerns
Good Typescript integration
A really good cli tool

And above all this good shit is the documentation. Of all the major JS fronted tools I would say that this one is the one i like working with the most all in account of how easy to find the shit that I need is. Have built some otherwise complex shit using nothing more than documentation....albeit i have done this with most frontend shit i use.

Me on a tech part of a job interview: "We shouldn't do it this way because it violates separation of concerns!"
Interviewer: *clearly impressed*

Me 2 weeks later: *does exactly what I said that shouldn't be done because I can't be arsed to do it the right way*

Ok, so our team is responsible for writing an app that consumes an API written by the client's team (I refuse to call it a "REST" API, despite their claims). On one of the clarification meetings we are discussing an endpoint that accepts a (logically) unique field multiple times, even though an entity is already registered in the system with that unique identifier. Our proposal would be that this API of theirs should not happily accept duplicates as many times as there are bits on a 4TB hard drive, rather it should signal an error.

The response we got is this: Due to the Separation of Concerns principle they thought that it should be our app's responsibility to not send a request if an entity with said field is already in the system. Thus there's no need for the backend to validate this.

I didn't hear the next part, because I had to collect my headphones from the other side of the room where they were flung in rage.

daily.
me: i looked into the customer dev's project and even though it's C#, i can use it as a source of inspiration for my own C++ library.
PM: okay, maybe we can even still use it, so that you use a C# dll with your C++ code.
me: ...
other colleague: that's a bad idea. it can already be a challenge to use unmanaged c++ in dotnet, but the other way round it's even more difficult. C# and C++ are languages that behave quite differently and it will be hard to implement a correctly working interface.
PM: okay. well... then please analyze this project's complexity in terms of LOC and create a class diagram, so we get an idea of how complex it is.
me: sure.
PM: hmm... maybe we should split this topic. since dev x will also rely on your library, analyze this project together with him, each of you look at another part of the classes.
me: that's.... i think that's a bad idea. implementing this functionality in this library is my job, not of dev X. he won't be involved in implementing any of the funcionalities and for him, it shouldn't matter how this works.
PM: yeah, but since we are prototyping, maybe we should just violate the "separation of concerns" rule.
me (internally): (ノಠ益ಠ)ノ彡┻━┻

in the end i could convince him to do it my way, but for fuck's sake... when was the last time he actually succesfully implemented something? 🤦‍♀️

Code languages should stay within their designated file extensions! I'm having to work with SQL in .resx files at the moment! Then there's HTML which is like a skitty little travelling monkey that I've found blundering about in SQL strings before, let alone people trying to sneak it into .js files, which is one reason I chose Angular over React btw (and don't give me that bullshit of separations of concerns vs separation of technologies).

It is mostly trivial to work with separate files across languages and doing so gives you the benefits of that languages support in the text editor or IDE you are using. It becomes easier to work with, the behaviour is predictable and it is formatted in a readable manner, ultimately reducing the potential for mistakes. Stop trying to make me put everything in strings!!

So about 3 weeks ago I was laid off from my dream job due to corporate bullshit. From the feedback received since then it is clear that the company made a mistake hiring a brand new React dev while they really needed an experienced one. Because the consultants who were supposed to be weren't. And the other in-house front end dev was an elitist asshole. And I never received proper feedback until it was too late. Actually I still don't have proper feedback save for some vague stuff which really sounds like the kind of feedback you'd give someone in the middle of their learning process. They even said eventually given more time I could have made it. But alas they felt they had to make a call in the best interest of the company.

Things moved fast since then, I took a week to recover and then I spent time updating my resume before getting back in touch with the recruiter who got me my last job. Great guy and he was happy to help me again. Applied to some positions, got some replies, first in person interview I go to they are immediately willing to take me on.

So now I'm supposed to start tomorrow but somehow I'm having my doubts. The company isn't an IT company but rather a fashion company. They believe in developing in house tools because past attempts with external companies resulted in them trying to push their vision through. Knowing who they worked with I agree, they tried to oversell all the time. But after talking with their developers I noticed they are behind on their knowledge. But so am I. So there was no tech interview which means I am getting an easy way in. And if they honour their word I'll be signing tomorrow for around my old wages.

So you'd think that sounds good right? And yet I'm worried it's going to be another shit show working on software without proper analysis or best practices. I mean the devs aren't total idiots, they are mediors like me and I think their heart is in the right place. They want to develop a good project but it will be just us 3 making a modern .net wpf application with the same functionality of the old Access based system currently in use. I was urged by the boss to draw on my experience and I think he wants me to help teach them too. But I'm painfully aware for my decade since graduating I'm a less than average .net dev who struggles with theory and never worked a job where I had someone more experienced to teach me. I coasted most of the time in underpaid jobs due to various reasons. But I'd always get mad over shitty code and practices. Which I realize is hypocritical for someone who couldn't explain what a singleton class is or who still fails at separation of concerns.

So yeah my question for the hivemind is what advice would you give a dev like me? I honestly dislike how poor I perform but it often feels like an insurmountable climb, and being over 30 makes it even more depressing. On the other hand I know I should feel blessed to find a workplace who seems to genuinely believe that people grow and develop and wishes to support me in this. Part of me thinks I should just go in, relax, but also learn till I'm there where I want to be and see if these people are open to improving with me. But part of me also feels I'm rushing into this, picking the first best offer, and it sure feels like a step backwards somehow. And that then makes me feel like an ugly ungrateful person who deserves her bad luck because she expects of others what she can't even do herself :(

So we had this legacy Objective-C codebase for a mobile app that was actually pretty good: I'd inherited the codebase and spent the past several years gradually improving it and I was actually quite proud of the work I put into it. So of course management decides to scrap it (with NO consultation from the engineers) and outsource a complete rewrite of the app in C# for Windows Universal.

Let me tell you. That code was without a doubt and without exaggeration the *worst* code I've seen in my close to 30 years of experience as a developer. I mean they broke every rule in the book, I'm talking rookie mistakes. Copypasta everywhere, no consistent separation of concerns, and yet way too many layers. Unnecessary layers. Layers for the sake of layers. There was en entire abstraction layer complete with a replicated version of every single data class *just* to map properties in pascal case to the same property in camel case. Adding a new field to a payload in the API amounted to hours of work and about eight different files that needed to be modified. It was a complete nightmare. This was supposed to be a thin client, yet it had a complete client-side Sqlite database with its own custom schema (oh and of course a layer for that!) completely unrelated to the serverside schema, just for kicks. The project was broken up into about eight or nine different subprojects, each having their own specific dependencies on various of the other subprojects in such a tightly-knit way that it made gradual refactoring almost impossible. This architecture was so impressively bad, it was actually self-preserving!

Suffice it to say it was a complete nightmare, and was one of the main reasons I ended up leaving that company. So just sayin', legacy code isn't always bad. :)

Separation of concerns is a beautiful thing.

JSX is fucking ugly. Fuck that shit. I hate JSX with a passion.

Here is one. Did you know that the digestive system works really hard to digest the food eaten?

How about we blend all the food before consuming it? Take a blender and add a cup of coffee, add some salad, add a piece of cake, a few slices of pizza, hot sauce and for good measure add some juice, or whatever-you-eat-for-lunch.

After all, all that food is going to get mixed anyway. This is more efficient!

No? Why not? Because it's ugly, highly unappetizing, disgusting even, and it takes away the pleasure of eating, the enjoyment of a good meal.

That in a nutshell is JSX: mashing up everything together under the pretext of efficiency.

Web development not only is an art, but above all must be enjoyable to those who devote their lives to it. And ugly ain't gonna cut it.

I'm the lead dev on this team. The project is split into multiple separate modules to comply with separation of concerns, and so new devs don't need the whole fucking codebase (risking them running away with everything) to contribute to the project as a whole.

So we don't need a fucking config file to enable and disable features.

So we don't need to upload a 500mb monolith every time we want to test a change.

So we can test old fucking versions of modules without merging it back into the entire codebase.

What did this fucking dev do? He was having one small issue with Maven. One. It wasn't updating his local snapshots to the correct Artifactory version.

He decided, instead of trying to fucking fix it: HEY, LETS IGNORE THE LEAD DEV'S DEMAND TO KEEP THEM SEPARATE. IM GOING TO MERGE THEM INTO ONE MODULE FOR SOME FUCKING REASON.

I refuse to continue working with this dev if he's going to sidestep my demands and undermine my authority. He wants to go it alone? Be my fucking guest. I'm not touching his shitty single-codebase monolithic monstrosity.

If this is going to be a regular fucking occurrence, he can eat a dick and choke on it.

Started taking an Angular 5 tutorial to see how things were going in the world of Angular JS. I got to say, I am impressed. It makes me think of React in a lot of ways, but with a heavy emphasis on separation of concerns. Particularly suited for those that do not like to mix views with logic. I am liking it and going at it with an open mind although React is still my preferred option. One thing that irritates me is the ammount of "plz sir, can you give code for <insert complex and heavy app that people just do not give for free>".....so annoying.

On another note, I like how Angular brings in the concept of di among other things to the table, what I am trying to get is the feeling of writing 2 apps, there is one thing to have MVC on the background, the other is to have it in the frontend! Oh well, Angular (first edition) was fun and I know it decently well, time to get cracking on more code!!

This started as an update to my cover story for my Linked In profile, but as I got into a groove writing it, it turned into something more, but I’m not really sure what exactly. It maybe gets a little preachy towards the end so I’m not sure if I want to use it on LI but I figure it might be appreciated here:

In my IT career of nearly 20 years, I have worked on a very wide range of projects. I have worked on everything from mobile apps (both Adroid and iOS) to eCommerce to document management to CMS. I have such a broad technical background that if I am unfamiliar with any technology, there is a very good chance I can pick it up and run with it in a very short timespan.

If you think of the value that team members add to the team as a whole in mathematical terms, you have adders and you have subtractors. I am neither. I am a multiplier. I enjoy coaching, leading and architecture, but I don’t ever want to get out of the code entirely.

For the last 9 years, I have functioned as a technical team lead on a variety of highly successful and highly productive teams. As far as team leads go, I tend to be a bit more hands on. Generally, I manage to actively develop code about 25% of the time to keep my skills sharp and have a clear understanding of my team’s codebase.

Beyond that I also like to review as much of the code coming into the codebase as practical. I do this for 3 reasons. I do this because as a team lead, I am ultimately the one responsible for the quality and stability of the codebase. This also allows me to keep a finger on the pulse of the team, so that I have a better idea of who is struggling and who is outperforming. Finally, I recognize that my way may not necessarily be the best way to do something and I am perfectly willing to admit the same. I have learned just as much if not more by reviewing the work of others than having someone else review my own.

It has been said that if you find a job you love, you’ll never work a day in your life. This describes my relationship with software development perfectly. I have known that I would be writing software in some capacity for a living since I wrote my first “hello world” program in BASIC in the third grade.

I don’t like the term programmer because it has a sense of impersonality to it. I tolerate the title Software Developer, because it’s the industry standard. Personally, I prefer Software Craftsman to any other current vernacular for those that sling code for a living.

All too often is our work compiled into binary form, both literally and figuratively. Our users take for granted the fact that an app “just works”, without thinking about the proper use of layers of abstraction and separation of concerns, Gang of Four design patterns or why an abstract class was used instead of an interface. Take a look at any mediocre app’s review distribution in the App Store. You will inevitably see an inverse bell curve. Lot’s of 4’s and 5’s and lots of (but hopefully not as many) 1’s and not much in the middle. This leads one to believe that even given the subjective nature of a 5 star scale, users still look at things in terms of either “this app works for me” or “this one doesn’t”. It’s all still 1’s and 0’s.

Even as a contributor to many open source projects myself, I’ll be the first to admit that have never sat down and cracked open the Spring Framework to truly appreciate the work that has been poured into it. Yet, when I’m in backend mode, I’m working with Spring nearly every single day.

The moniker Software Craftsman helps to convey the fact that I put my heart and soul into every line of code that I or a member of my team write. An API contract isn’t just well designed or not. Some are better designed than others. Some are better documented than others. Despite the fact that the end result of our work is literally just a bunch of 1’s and 0’s, computer science is not an exact science at all. Anyone who has ever taken 200 lines of Java code and reduced it to less than 50 lines of reactive Kotlin, anyone who has ever hit that Utopia of 100% unit test coverage in a class, or anyone who can actually read that 2-line Perl implementation of the RSA algorithm understands this simple truth. Software development is an art form. I am a Software Craftsman.

#wk171

One responsibility of our team is general code QA for the entire dev department, DevMgr walks in our area yesterday…
DevMgr: “Has anyone reviewed the new WPF threaded model execution code?”
- everyone on the team responds “no”
DevMgr: “Can we get a review on that code ASAP? If it works as well as the developer said, it’s going to solve the lock up problems users are experiencing and automatic logging of errors.”
DevA: “Well, no amount of code is going to stop users from performing bad searches locking up the user-interface. That code is just a band-aid around the real problem. If the developers would write unit tests first …”
- rant about 5 minutes on unit testing that had nothing to do with why the DevMgr was here
DevB: “Yea, the code probably isn’t written to handle threads correctly. All the threading they’ve done so far is –bleep-”
DevMgr: “Oh, I wasn’t aware of that. Get me the results of the code review and if they don’t have unit tests, delete it from source control and let the developer know it’s not up to our standards.”
OMFG!! You have not even seen the code!
OK, DevA ..what the –bleep- does unit testing have anything to do with the user interface! You know the DevMgr is too dim to understand the separation of concerns. Shut your pompous ‘know-it-all’ mouth.
DevB…what the –bleep- have ever done in WPF? You manage the source control and haven’t written any C# in two years and never, ever written code for any significant project. Take that “handle threads correctly” and shove it up your –bleep-. Pompous –bleep-hole. Go back and watch youtube and read your twitter while the grown-ups get the work done.

I'm starting to think that DRY is overrated.

The only thing that I think works great in Node.js ecosystem is Socket.io

Otherwise anything JavaScript related is too bad for me. So many frameworks releasing each month. First it was React then people said that vue is better... Now hearing Svelte is the best. This shit is going crazy.

Personally I prefer to keep back end in a different language such as PHP or Python. Separation of concerns was a thing some years ago now everything is JS.

Are there other alternatives to Socket.io in other languages which are easy to setup just like Socket.io? XMPP is there but I feel it is overly complicated to get started.

A sweaty furry sodomizing a dead dog would still be less disgusting than the codebase on which I have to work, some highlights are:
- The same class repeated 40 times with little variations instead of using some decent parametrization
- Inexistent encapsulation and separation of concerns, most changes requires to modify and recompile 2-3 indipendent Maven projects
- Abuse of inheritance which instead of being used to create "is-a" relationship as it should be it's used to reuse some methods of a class in another instead of using Spring dependency injection as we should be

It would be understandable in a 20 years old legacy projects but in something which started 2 months ago it drives me mad, I tried to fight to change it but in the big enterprise to which I'm "body-rented" it's impossible

how do you go about working with a teammate that the best thing would be to rewrite the code he worked with for the last 5 months!?

I mean, no separation of concerns, layers upon layers of unnecessary abstractions, unneeded parallelism and mutex and whatnot...

Like, 5000+ lines of code that could be done in 400...

Let me rant! I don’t usually do this but this is just frustrating and draining. Please tell me if im wrong. We have authentication that needs to be refactored. I was assigned on this issue. Im a junior btw. I also attached an image of my proposals. The issue of the old way of our signup process is that when validation fails they will keep on accepting the TaC (terms and conditions) and on our create method we have the validation and creating the user. Basically if User.create(user_params) create else throw invalid end. (Imma take a photo later and show it you)which needs to be refactored. So I created a proposal 1. On my first proposal I could create a middleware to check if the body is correct or valid if its valid show the TaCs and if they accept thats the moment the user is created. There is also additional delete user because DoE told me that we dont need middlewares we have before and after hooks! (I wanted to puke here clearly he doesn’t understand the request and response cycle and separation of concerns) anyway, so if middleware is not accepted then i have to delete the user if they dont accept the TaCs. Proposal 2. If they dont want me to touch the create method i could just show the TaCs and if they dont accept then redirect if they do then show form and do the sign process.

This whats weird (weird because he has a lot of experience and has master or phd) he proposes to create a method called validate (this method is in the same controller as the create, i think hes thinking about hooks) call it first and if it fails then response with error and dont save user, heres the a weird part again he wants me to manually check on each entity. Like User.find_by_email(bs@g.com) something like that and on my mind wtf. Isnt it the same as User.create(user_params) because this will return false if paras are invalid?? (I might be wrong here)

This is not the first time though He proposes solutions that are complex, inefficient, unmaintainable. And i think he doesnt understand ruby on rails or webdev in particular. This the first time i complained or I never complained because im thinking im just a junior and he hs more experience and has a higher degree. This is mot the case here though. I guess not all person who has a higher degree are right. To all self thought and bachelors im telling you not all people who went to prestige university and has a higher degree are correct and right all the time. Anyway ill continue later and do what he says. Let me know if im wrong please. Thanks

jinja templates make me look towards html in a whole new light. are we 'inserting' data to an already rendered page? am i really mixing server code with ui ? It doesn't feel so. there are if else and loops being executed for html code, like wtf?

I don't know but everything feels so good. like i was literally hating every piece of website i was writing in php. everytime i wrote <div>....</div> followed by <?php ... ?> followed by another html tag /php tag in a fuckin php file, i wanted to kill someone from w3c.
WHY THE FUCK ARE WE ALLOWING THE MIXUP ?WHY IS PHP FILE HOLDING HTM TAGS? WHY?WHY?WHY?

But this... this is beauty. their is separation of concerns. jinja has some big powers, we can loop, repeat, make clauses, inherit other html classes, load html content into blocks, set variables,
but main concepts like file handling, response/request handling,calculations,etc are all being done in separate python files. I know that these jinja templates also might be running python in background, but atleast a developer cannot fuck up that code.
we can be sure that if correct jinja codes are written in html, then it would load correctly. And wherever devs doesn't fuck up, the output is better to understand and more maintainable/scaleable

I tend to overengineer. Why? Because I had a view in JavaFX with its controller that had a bunch of key listeners which changed the UI. I wanted to change the view based on wifi connection/no connection with a server, which was managed in a Client class. The controller took the client to give it a message that client then had to send. For "separation of concerns" I created a separate view + controller for the "not connected" state.
Now the Client knew all about the connection, so I put up the Observer pattern and wanted the Main (Application) class to swap the layouts as an Observer of the Client. After an Exception on FX thread and Platform.runLater(), to solve the issue, I faced a new problem: the key presses weren't executed anymore. I still don't know why this happens. Maybe I'm missing something.🤷
Then met with one of my group partners (it's a uni project):
Let's attach the Observer to the original controller. Have only the original view that changes due to the controller updates as Observer. Let's see if that might even remotely work...🤔
It worked🤦😂

So they develop this app. That uses our front end component library. That queries a GraphQL layer developed as NPM package. That uses a data service abstraction NPM package. That uses another NPM package mapper library. That queries an old REST API returning XML.

It takes days to make a newly added XML node in the bottom-most layer available in the app, requiring changes to 4 repositories and 3 NPM releases.
Refactoring is dead, because 1 change will affect all layers. And the worst part is: theres only 1 app using these packages, so no case for re-use. Overzealous separation of concerns I guess?

I hate heavy-duty frameworks. I'm currently working with both ABP and Angular and they're both just godawful. They're inelegant solutions that just fucking eat through my computer's RAM. It's just slow, generates code that my IDE analyses and finds problems with. And the application runs so. fucking. slowly. I like how it enforces a separation of concerns, but that's about it. Oh and in the name of God, why would anyone do anything with angular now that we have Vue? Why? I can't...

As an android dev when I inherited a shitty project thats when I realized what really means to write readable and most importantly testable code. Codebase I inherited wasnt even really that bad it was quite readable, but boy it was not suited for any unit/instrumented tests. im talking spaghetti code.

Nowadays I refactor apps to make sure they are testable instead of spending weeks writing tests for a shitty codebase which was done without thinking about separation of concerns. Clients hate the extra couple weeks on top of request but what can I do, if they want tests they need to work with TDD approach or give extra time for refactors.