We're using a ticket system at work that a local company wrote specifically for IT-support companies. It's missing so many (to us) essential features that they flat out ignored the feature requests for. I started dissecting their front-end code to find ways to get the site to do what we want and find a lot of ugly code.
Stuff like if(!confirm("blablabla") == false) and whole JavaScript libraries just to perform one task in one page that are loaded on every page you visit, complaining in the js console that they are loaded in the wrong order. It also uses a websocket on a completely arbitrary port making it impossible to work with it if you are on a restricted wifi. They flat out lie about their customers not wanting an offline app even though their communications platform on which they got asked this question once again got swarmed with big customers disagreeing as the mobile perofrmance and design of the mobile webpage is just atrocious.

So i dig farther and farthee adding all the features we want into a userscript with a beat little 'custom namespace' i make pretty good progress until i find a site that does asynchronous loading of its subpages all of a sudden. They never do that anywhere else. Injecting code into the overcomolicated jQuery mess that they call code is impossible to me, so i track changes via a mutationObserver (awesome stuff for userscripts, never heard of it before) and get that running too.

The userscript got such a volume of functions in such a short time that my boss even used it to demonstrate to them what we want and asked them why they couldn't do it in a reasonable timeframe.

All in all I'm pretty proud if the script, but i hate that software companies that write such a mess of code in different coding styles all over the place even get a foot into the door.

And that's just the code part: They very veeeery often just break stuff in updates that then require multiple hotfixes throughout the day after we complain about it. These errors even go so far to break functionality completely or just throw 500s in our face. It really gives you the impression that they are not testing that thing at all.

And the worst: They actively encourage their trainees to write as much code as possible to get paid more than their contract says, so of course they just break stuff all the time to write as much as possible.
Where did i get that information you ask? They state it on ther fucking career page!

We also have reverse proxy in front of that page that manages the HTTPS encryption and Let's Encrypt renewal. Guess what: They internally check if the certificate on the machine is valid and the system refuses to work if it isn't. How do you upload a certificate to the system you asked? You don't! You have to mail it to them for them to SSH into the system and install it manually. When will that be possible you ask? SOON™.
At least after a while i got them to just disable the 'feature'.

While we are at 'features' (sorry for the bad structure): They have this genius 'smart redirect' feature that is supposed to throw you right back where you were once you're done editing something. Brilliant idea, how do they do it? Using a callback libk like everyone else? Noooo. A serverside database entry that only gets correctly updated half of the time. So while multitasking in multiple tabs because the performance of that thing almost forces you to makes it a whole lot worse you are not protected from it if you don't. Example: you did work on ticket A and save that. You get redirected to ticket B you worked on this morning even though its fucking 5 o' clock in the evening. So of course you get confused over wherever you selected the right ticket to begin with. So you have to check that almost everytime.

Alright, rant over.
Let's see if i beed to make another one after their big 'all feature requests on hold, UI redesign, everything will be fixed and much better'-update.

At one of my former jobs, I had a four-day-week. I remember once being called on my free Friday by an agitated colleague of mine arguing that I crashed the entire application on the staging environment and I shall fix it that very day.

I refused. It was my free day after all and I had made plans. Yet I told him: OK, I take a look at it in Sunday and see what all the fuzz is all about. Because I honestly could fathom what big issue I could have caused.

On that Sunday, I realized that the feature I implemented worked as expected. And it took me two minutes to realize the problem: It was a minor thing, as it so often is: If the user was not logged in, instead of a user object, null got passed somewhere and boom -- 500 error screen. Some older feature broke due to some of my changes and I never noticed it as while I was developing I was always in a logged in state and I never bothered to test that feature as I assumed it working. Only my boss was not logged in when testing on the stage environment, and so he ran into it.

So what really pushed my buttons was:

It was not a bug. It was a regression.

Why is that distinction important?

My boss tried to guilt me into admitting that I did not deliver quality software. Yet he was the one explicitly forbidding me to write tests for that software. Well, this is what you get then! You pay in the long run by strange bugs, hotfixes, and annoyed developers. I salute you! :/

Yet I did not fix the bug right away. I could have. It would have just taken me just another two minutes again. Yet for once, instead of doing it quickly, I did it right: I, albeit unfamiliar with writing tests, searched for a way to write a test for that case. It came not easy for me as I was not accustomed to writing tests, and the solution I came up with a functional test not that ideal, as it required certain content to be in the database. But in the end, it worked good enough: I had a failing test. And then I made it pass again. That made the whole ordeal worthwhile to me. (Also the realization that that very Sunday, alone in that office, was one of the most productive since a long while really made me reflect my job choice.)

At the following Monday I just entered the office for the stand-up to declare that I fixed the regression and that I won't take responsibility for that crash on the staging environment. If you don't let me write test, don't expect me to test the entire application again and again. I don't want to ensure that the existing software doesn't break. That's what tests are for. Don't try to blame me for not having tests on critical infrastructure. And that's all I did on Monday. I have a policy to not do long hours, and when I do due to an "emergency", I will get my free time back another day. And so I went home that Monday right after the stand-up.

Do I even need to spell it out that I made a requirement for my next job to have a culture that requires testing? I did, and never looked back and I grew a lot as a developer.

I have familiarized myself with both the wonderful world of unit and acceptance testing. And deploying suddenly becomes cheap and easy. Sure, there sometimes are problems. But almost always they are related to infrastructure and not the underlying code base. (And yeah, sometimes you have randomly failing tests, but that's for another rant.)

TFW your client's git policies are so draconian that the dev teams use "develop" as trunk, and completely ignore the release process.

I wrote up 50 pages of git standards, documentation and procedure for a client. Bad indian director 9000 decides the admin (also Indian) who specializes in Clearcase and has no git or development experience is more qualified to decide and let's him set the policy.

FF to today:

- documentation, mostly contradictory, is copy pasted from the atlassian wiki
- source tree is the standard
- no force pushing of any branches, including work branches
- no ff-merge
- no rebasing allowed
- no ssh, because he couldn't figure it out...errr it's "insecure"
- all repos have random abbreviated names that are unintelligible
- gitflow, but with pull requests and no trust
- only project managers can delete a branch
- long lived feature branches
- only projects managers can conduct code reviews
- hotfixes must be based off develop
- hotfixes must go in the normal release cycle
- releases involve creating a ticket to have an admin create a release branch from your branch, creating a second ticket to stage the PR, a third ticket to review the PR (because only admins can approve release PRs), and a fourth ticket to merge it in
- rollbacks require director signoff
- at the end of each project the repo must be handed to the admin on a burned CD for "archiving"

And so no one actually uses the official release process, and just does releases out of dev. If you're wondering if IBM sucks, the answer is more than you can possibly imagine.

Ok. so my ISPs port 53 was DDoSed and I couldnt resolve hostnames anymore.

The only app that worked was Telegram because it (thankfully) calls an ip and doesnt use a DN.

The friends at @codrTalk were so kind to resolve like 20 sites for me by hand in order for me too 1) find out what happened 2) find hotfixes 3) download tor

Now, that I downloaded tor, im doing this over tor.

PWEASE @devrantron team and @devRant team

Could you please make it, that devRant get called over ip as opposed to by DN in your apps?

Anyway, thanks to those in codrTalk.

There is a company providing a very speciffic service. And it has a core application for that svc, supported by a core app team.
That company also has other services, which are derivations of the core one. So every svc depends on core.

Now that we're clear on that... I was working in a team of one of the subservices. We very strongly depended on core. In fact, our svc was useless if integration w/ core broke down.

The core team had an annoying habbit. They refused to version their webservices and they LOVED to push api updates w/o any warnings. Our prod, test, other envs used to fail bcz of core api changes quite often. Mgmt, IT head was aware of the problem and customers' complaints as well.

So as a result, once core api changes we're all in a panic mode: all prior priorities are lowered and revival of prod is to be our main focus. Core api is not docummented, the changes are not clear, so we have to reverse engineer the shit out of it. We manage to patch our prod up w/ hotfixes, but now we have tech debt. While working on the debt, core api changed again, in test env. Mgmt pushes debt back and reallocates us to hotfix test. Hotfix is 80% done when another core api breaks. Now mgmt asks us to drop wtv we're working on and fix that new break. By the time we're to deploy the hotfix, another api breaks in another env. The mgmt..... You get the picture :)

2 years go by, nothing has changed so far.

When after 6 hotfixes later the bug is still present:

This week I learned:

a.) Always quadruple check new feature before pushing a new update - otherwise it will end with one to three hotfixes
b.) Never do the last minute changes to the features - that ends with two to four hotfixes
c.) Always make sure that the update system works - otherwise no one will get your hotfixes

Geez, how many hotfixes can we push in a week?

getting part time job as a "junior web developer" while doing my uni, things go well at first. Now my profession is, senior web developer ... and database administrator, and server admin, and the one who call for hotfixes, and who code js for frontend, while under paid because I am still "a college student" ..

Today I got a reply to a PR comment from a coworker literally stating that "it's not a good practice to add unit tests to hotfixes".

I can't, just can't. Left me speechless. Wonder where this guy gets his "good practices" from.

my company is "pivoting" way too goddamn fast; they are pulling devs from other projects and throwing them into something that is a fragile system (was supposed to be replaced already) and is using a completely different stack than most of them usually work with. they keep promising 3rd parties that we will knock out their requirements within a week or two, and as such they are pushing haphazardly merged feature branches into production with absolutely no regression testing.

then when shit explodes and operations grinds to a halt, they tell the half of the team that actually knows how the system works to drop everything and fix it, and leave the diverted devs to continue to develop shit based on requirements drawn on a cocktail napkin, and then they fucking push both the hotfixes and the newest features at the same time.

I probably have the most tribal knowledge at this company, and they are paying me ok, so it's enough for me to just pour some rye and suck it up for the time being and milk the gig. but this can't be sustainable, right? i'm passively looking around for other work, as I've already had enough being here for over 3 years, but i'm finding most places to be slow on the application/hiring process lately due to covid.

Edit: i think i used the wrong tag here, but what the fuck ever. i haven't figured out how to tag shit properly on any platform

Extremely frustrated with the release process and versioning system at my current company. Don't know if this is same everywhere or the half ass release managers can't think of a better way here.

Basically for any client raised issue that can't wait for next release are built as a hotfix. However hotfixes are never bundled togather or shiped to other clients. This is causing a vicious chain, two clients raise two separate issues on same version. Instead of fixing them as single hotfix (however minor the issues) we create two hotfix versions for each with only their issue. A week later same clients come back with the issue the other raised. Once again instead of bundling what is now effectively same code we build hotfixes on top of the clients respective branches. We now have two branches to maintain with same codebase. No matter how serious issue, the hotfix is never made generally available and always created on client's specific hotfix version.

Now that was an example for only two clients, in reality we have released five patch versions of a product in last 2 years. Each product version contains about a dozen artifacts (webapps, thick clients, etc) with its own version. Each product version being shipped to various clients. Clients being big banks never take a patch of product even if it fixes their issues and continues requesting hotfix. We continue building hotfixes on client branch and creat ever increasing tech debt. There is never a chance to clean up or new development. Just keep doing hotfix after hotfix of same things.

To top if all off, old branches are still in svn while new in git. Old branches still compile with ant new with maven. Old still build with java 5,6,7 while current with 8. Old still build from old jenkins serve pipelines while new has different build server. Old branches had hardcoded integration db details which no longer exists so if tou forget to change before releasing it doesn't work.

Please tell me this is not normal and that there are better ways to do this? Apologies I think I rambled on for too long 😅

Just finished coding/support/putting fires down marathon that lasted 2 weeks.

Basically I launched 2 gaming servers got around 600 online players and had to spend 2 weeks around 12 hours everyday supporting them doing hotfixes and etc. I had no time for life. Ioved the work, debugging, responsibility and making impact. Also made some money.

Now I feel completely drained, sick and tired. I think I hit a burnout. I thought I will enjoy my time off with my gf but I just wanna lie in bed and dont do shit for at least 2-3 days. Extended stress over time did a number on me, I am now a nervous wreck.

How can I bounce back asap after a burnout? I have 2 weeks to rest and will have to launch 2 projects again and deal with christmas/new year bullshit...

Oh here's a good one. When the managers realised one of our apps is a giant hunk of crap that wasnt thought through at all and was lazily thrown together, and their solution is "meh let's just rewrite it in Swift on our new platform. And those other guys can maintain the old one and continue to do hotfixes for it until we are done".
I've been telling them for the past year that its the worst codebase I have ever seen and the lack of tests is disgusting and not something we should dare to release to paying customers (especially when those customers work in healthcare!!!). The best part was when one of them promised we would all be working on the new shiny platform by Christmas. That was last year. And I'm currently the poor bugger doing the legacy maintenance and in the process of trying to get moved to a new project. So much for managers promises amirite...

Spent most of this week busting my ass working on a hotfix that came out of nowhere with mega high priority. This annoys me greatly because the hotfix wasn't even fixing a bug, it was adding new functionality because certain customers were being blocked from testing without this specific feature. In my humble opinion, given that we release every weekend, hotfixes should be reserved for actual critical bugs. But anyway, as I probably could have predicted, the code got to QA and exploded. Literally nothing works.
This is what happens when you try to rush out features to satisfy customers. If you try to rush something that is late, you WILL make it later.
Meanwhile there's an issue I'm supposed to be fixing for our next release which goes out this weekend and I've had no time to even look because of this hotfix. And now it's the end of the day and I just feel worn out from stress, tomorrow will no doubt be similar.

You know what I hate? The periods close to a new release. It's that annoying time that managers start heaving impatiently to developers: "Push out new hotfixes! Push out new hotfixes!". Fuck you, man. Don't stress a developer. Kthx. If you think developing is so easy, come do it yourself, you IT-ignorant buffoon.

What the hell kind of tool is Gitlab? I just want to automatically backmerge hotfixes from master to development. Even fucking Bitbucket had a checkbox to enable this. But not Gitlab, no, you better create a pipeline job in your already unreadable, overcrowded pipeline yml, but oh, the checked out repo in the pipeline is a detached head and you cant push with the user that checks out there. So what, just use a project acess token which revokes after a year breaking your task and then switch origin amd branch manually. But your token-user can't push to protected branches, so create a merge request instead, which requires approvals, making the automated step no longer automated.
But dont worry, you can just use the gitlab api to overwrite the approval rules for this MR so it requires 0 approvals. But to do so you must allow everyone to be able to overwrite approval rules therefor compromising security.
And so you made a feature that should effectively be a checkbox a 40+ line CI job which compromises your repo security.

which nuthead of an architect is responsible for the way gitlab (and its CI) is designed?

Good times: Migrating a Jenkins build pipeline patched together out of groovy, python, bash, awk, perl scripts and God knows what else since I have only scratched the surface so far, from Maven to Gradle while not breaking day-to-day builds, integrations and deployments of features, hotfixes and releases. I'm actually enjoying the challenge but it's taking forever due to several issues:
- Jenkins breaks/hangs randomly because it's Jenkins
- Gradle can't handle sets of version ranges but Maven can
- Maven can't handle Gradle style version ranges
- Gradle doesn't have a concept of parent poms, you need to write a plugin and apply stuff programmatically. But plug-ins being part of the buildscript{} don't fall under depency management rules :clap:
- Meme incompatibility issues of BSD vs GNU versions of CLI tools like sed, grep etc

How do you deal with situation when u need to merge multiple feature branches to develop branch? All feature branches have develop branch as base. So as soon as one feature is merged to develop then there will be contlicts in other feature branches.

Should I merge first feature branch to develop, then rebase feature2 branch on most recent develop and then merge the rebased feature2 branch to most recent develop and continue like that?

What if later I need to do hotfixes for previously merged branches? Should I revert them then rebase them on most recent develop and once again merge it to develop? Or should I just make small commits for fixes.

Hasn't technology become magic - and we ignorant sorcerer's apprentices who can perform a video call to the other side of the globe perhaps while we understand only some bytes of the thousands software snippets that were piled up by us code monkeys to perform the miracle? ...

This however has always been the state of software (for us developers): that this house of cards needs constant care by our hands to not collapse - in constant fear we may preserve the facades while the number of components that interact, the sheer mass of code only allow for guesswork and hotfixes accumulating the technical debt. Yes, we have all that terms for that. The problems are known since the 80s or 60s, so we might be relabeling it once in a while, but mainly it is just: complexity.. or entropy.

Has anyone had success with GitFlow hotfixes and GitHub branch protection rules? Finishing a hotfix requires pushing directly to develop, but GitHub prevents it if PR policies are set up :/

This was my first week as Jr Dev. I'm excited, first commits, PR, issues and hotfixes. Any advice for me?
Have a nice weekend!!