Erlkönig: Complaints

Almost everything we use on the Internet is based on projects begun in Unix, or started and/or still running in some descendent like Linux, Android, OS X, etc. Many modern networked games played in consumer operating systems are in worlds that run on Linux hosts inside the company.

Microsoft, in contrast, was bad at all of these things, with abysmal implementations of networking (we had to use Netscape's special Internet handler because MS's was so bad for years), local-only email (MS thought companies would only want email internally), lackluster file sharing, a horrible window system (3.1, despite better ones being widespread in Unix variants at the time), an OS pitiabily focused on a single-user per box, and any number of other aspects of computing. Its one actual power is marketing.

For better perspective, consider: If Windows were to die one day (possible since it can be centrally killed from headquarters) it will take out a lot of companies and trouble a lot of individuals. However, if Linux disappeared (which can't be orchestrated by a central company) most of the Internet vanishes, and a lot of companies - including many the customers thought were Windows only - will die, while a smaller group of individuals will be troubled for a shorter period of time (since they can switch to another Unix flavor, or build a new one).

• Reboots without warning, often to do system updates at the worst times
• Often forcibly delays shutdown, regardless of real-world situations
• Often forcibly delays boot, regardless of real-world situations
• Fails at the most basic things, like finding ethernet or halting
• Gives abysmal feedback, often not even that inane spinner
• Generally useless help documentation
• Idiotically posits Reboot as a subcase of PowerOff (which would prevent rebooting)
• Designed to take consumers' money rather than to empower them
• Extensively spies on the user
• Insists the end user is unworthy of control of the hardware
• Unlikely to run for long before failing (compare to unix systems often running uninterrupted for years)
• Can't move windows that are busy, or more generally stupidly relies on the app running window to react to basic operations like moving a window
• Can't type into a window while keeping most of it under another one
• Popups frequently steal mouse input without warning
• Can't move, resize, or scroll most parent windows of popups
• More recent UI requires trial and error to find a grabbable part of a window to move it (unless using the shortcut)
• Apps frequently crash without describing why (other than vomiting up an address) nor logging anything for the end user somewhere findable
• Is still terrible at multiuser anything

• Ubuntu "snap"s have numerous problems because the snap devs seem to be unaware/uncaring of how linux is used in large deployments, notably by failing to support homes outside of /home
• NetworkManager only introduces problems for static server IPs
• systemd has been and likely always will be a terrible architectural choice, doing far too much for an allegedly single service and deeply retarding development of competing, topic-specific services
• TTYs are superset of teletypewriters (minus the deforestation), X has terminals which are supersets of TTYs, Wayland is NOT a superset of X's core abilities (graphics over network) and therefore CANNOT replace it.
• In Linux, i.e. in the X window system, the cross-editor generic way to paste any currently highlighted text to somewhere else is just the middle mouse button. So copy+paste is left-mouse-swipe-to-highlight then middle-mouse-click (and note the X defaults to automatically changing input focus to wherever the mouse moves, so you don't have to click the target window just to make it active) You can add other shortcuts, like the stupidly long set of actions from Windows to do the same thing, but implement highlight-swipe + middle mouse FIRST.

• WASD is common, but ESDF is better, and your game had better support arbitrary key bindings, but whatever the devs choose, it'll probably be wrong.
• Your game may not be paying the end user, but if it crashes it's often worse than a work application crashing - be serious about making your game stable and resilient.

A number of facets of the real world are vastly less tidy than the average naïve developer wants to believe:

• NO! So you think all names are at least first + maybe-middle + last and made from ASCII letters, and that spaces and punctuation don't occur in them, and that each always starts with a capital letter?
• NO! So you thought real-world human names were unique?
• NO! So you think the street address + state + zip + country works worldwide? [in some areas, a drawing is required for success]
• NO! So you think there are always 60 seconds in a minute, 24 hours in a day, and daylight stupid time is always an offset of one hour and doesn't change year to year? [and historically, i.e. pre-railroad standardization, timezones were insane]

So, you think it's fine to just skip worrying about memory exhaustion, skip checking for 0 returns from malloc(), configure the system to overallocate memory to malloc will lie, and even write this same laissez-faire attitude into your libraries so any program that uses your library loses the ability to manage memory as well? I hate you. I watch programs die all the time because of this deliberate, avoidable blockheadness, and they frequently carry unique, unsaved user content to the void with them.

So SCRUM usually doesn't work (there are exceptions), and Kanban doesn't lie about whether goals were completed. So of course zillions of managers claim they're using SCRUM (aka. "sprints", although often merely modernity plastered over waterfull planning underneath). This allows them to evade Kanban's reliable truthfulness about progress.

The result is that the sprints initially complete on schedule, but then bog down as more and more patches have to be planned in to deal with the too-many shortcuts taken on earlier sprints to meet the deadlines. The unassigned priority one ticket pile grows without bound.