It's long, but trust me you won't get bored. So today, I went to work, even on a Sunday. My supervisor had given the task to finish off my custom module by this week.
There were only the 2 of us in the office. Everyone else enjoyed the holiday.
I got stuck somewhere in the code and approached his desk. We could never see what he was doing on his computer.
Suddenly, I could find him uneasy as I approached and he started jolting St those "ctrl+w" and "alt+f4s" and clutched his mouse to minimize. I said, "Uh sorry, sir but this taxonomy doesn't work in this code, help me out?"
He said, "Oh sure, sit. And he opened chrome trying to act cool. Guess what? Chrome played it like a boss, "Google Chrome didn't shut down correctly. Restore the following pages.
[] Shocking! 99 year old couple have s3x
[] xxx tiny teen shows her.... (i don't remember exactly.)"
The quickest possible glance I ever had. And the most sweaty face I ever saw of his.
He granted me a leave today and extended my deadline for the next week as well. I thought I was screwed :P

God I fucking hate WordPress

Today I found out a single WooCommerce site is using 20% of database CPU (for comparison, 50 other sites use only 10% between them). Upon investigation it turns out that WooCommerce loads every single product in order to do the taxonomy count. Because the data is stored as PHP-serialised values in post meta, there's no way to do a decent "SELECT count() ... GROUP BY ...". And that code came straight from the WordPress devs, no more and no less...

The joys of Drupal.

Coworker: i can't hide this option from this select. it's been created programmatically by another module.
Me: doesn't that mean you defined it somewhere?
C: nope, the module does it with the whole taxonomy
M: can't you interact with it somehow,
C: i don't know
M: you don't know?
C: it's a module by the community, i don't understand it
M: *what the fuck.* ...fine. can you hide it with some js for now?
C: i tried no. they get loaded at different times from different behaviours.
M: then... what?
C: i don't know, i've been searching drupal.org for the whole morning to no avail.
M: *god. damn.* Create the select with something else, then? maybe by raw php?
C: that's bullshit! you don't create elements lile that in 2018! no one writes php or html anymore, unless he"s retarded!

Bloody hell. I'm not covering for this. My part is done (in rails) and i'll deliver it this afternoon. Not for this kind of delay.

** few years ago **
Other team: "There's a lot of overlap and misclassification between issue categories A and B; We want to merge them into one in our taxonomy"
Me: These are very different issues and have very different root causes. Shouldn't we fix the definitions instead? We don’t have major drivers for B now but if we remove it, it might bite us in year or two.
** loads of back and forth, and eventually B gets killed

** year and a quarter passes
Executive meeting: “So you’re telling us we can’t properly quantify the impact of B because we can’t distinguish A from B. It looks very clear cut to me why don’t we have split for these already and how long will it take to implement it?”
Me: Biting my tongue not to throw anyone under a bus.

Is an art director/manager that makes designs for an existing website expected to brief the developer on how he wants his new designs to work? Or is the developer expected to innately understand what his functional desire is from the design itself?

My art director/manager claims that he shouldn't have to log into the backend to see how things relate to one another, and that his designs alone should imply what is intended.

When some design element overlaps an existing image gallery, am I expected to magically know whether this element is singular or partaining to the current image shown?

I want to know whether or not me getting mad at him for not telling me how he wants stuff to work is valid, and whether or not I should demand that he briefs me how his design relate to existing taxonomy.

Am I the bad guy?

When your php-code and the taxonomy of your site is perfect, but a plugin screws the breadcrumbs totally :(

Working a week on LSTM based text classifier, getting 89% accuracy only to then get better result with Logistic Regression which was supposed to serve as baseline, lol. Background: 180+ classes of google product categorization taxonomy, 20 million rows of data items (short texts). Had a similar experience once on sentiment classification, where SVMlight outperformed NN models.