We Get Letters Oct/Sep

By Michael W Lucas

Dear IDCblahblahblahwhatever,

I have previously discussed both new releases and packaging software in the pages of this very Journal, but I’m certain you recall none of that. Modern systems administrators have outsourced their memory to online forums and search engines, which worked until illiterate large language models got rebranded as AI and the resulting feral autocomplete engines stuffed your external brain’s technology section with reconstituted search-engine-optimized Scorpion fan fiction. Fortunately for you, so did your boss. You have the option to fall back on the computer’s built-in manual, whereas your boss thinks books like The Seven Highly Effective Cheese Mover Habits contain undying management wisdoms. He doesn’t remember anything from that either, but he keeps it on the back of the loo to present an image to anyone foolish enough to enter his lair. No, don’t touch it, the cheap paper those things are printed on absorb ambience and I should know.

But did you consider, possibly, for a moment, working on the legitimate issue underlying your question? No, you did not, as evidenced by the fact that whatever circuitous “reasoning” process that led you to write and send this letter betrayed you by permitting you to, once again, touch not just a computer but the Internet. Yes, yes, I am also on the Internet, but I am inoculated by my memories of using hardware with nine-and-a-half-bit bytes to send messages to email addresses optimized for teletype. My very bones know that this shambling horror will betray us, and any time my mere meat attempts something so foolish as to rely on digital resources they smash the offending limb into the closest relatively immovable object. We built the Internet as a repository for the master list of silly possum jokes, and greedy children ruined the whole thing. We predicted this failure mode, of course—not in precise detail, but it will all end in tears is pure prophecy.

You’ve been so foolish as to install an operating system.

Then you added more software to do things.

Presumably it works for sufficiently generous values of “works.”

As with so many systems administrators, you believe that the cure for your disease is more disease. Fine. Let’s run with that and see what deep damp pit you wind up in, and what kind of beetles you’ll spend the rest of your deliciously short life dining on before they return the favor.

“Custom built.” There. That’s your problem.

Did you write the software? Fine. If you had written it well, you wouldn’t be asking this question. No, I’m not insulting you. I’m offering a hand up to my level. I recently published a program I wrote for production use, ostensibly to demonstrate why you should not use my code. It was immediately declared “comically evil” and if everyone who sent me a refactored version had accompanied it with a dollar, I would not be writing this column but instead living my dream of being the first human in history to die of Gelato Degeneration. If you had written your code well, you would not have asked. Not because your code would work, but because you’d know it wouldn’t. A new release means new testing. Do it. (If you released your code to the public thinking someone might find it useful, you have done everyone a disservice. I released mine not only because the code itself was ghastly, but because it provided my publishing bibliography as an SNMP module and the resulting horror among people who understood what I had unleashed supported my long-term goal of making computing too repulsive for polite society.)

But probably you scrounged a few programs off the Internet. Software written by people that were not wise enough to keep their mistakes to themselves. Unlike me. You used pipes and redirects to glue these tiny atrocities together, the way sysadmins have done ever since Thompson and Ritchie declared their first Unix system beta-ready and offered an account to a soon-to-be-ex-friend. Your manager took you seriously when you said that everything worked, and now that you’ve tied yourself to the tracks, the upgrade trolley’s coming and you’d like me to throw the switch to divert it onto your boss who’s tied himself to a different track. I categorically refuse. Partly because I firmly believe in that most precious and fundamental of rights, the right to take the consequences, but mostly because I have sufficient trolleys for everyone.

Perhaps you (ugh) bought custom software. I can’t help you, but another purchase might. Might.

Wherever it came from, custom software causes misery. What do we do with misery?

That’s right. We share it.

The ports system exists to not only share misery, but to reliably replicate it across hundreds or thousands of users. (I know the documentation doesn’t state that, but it certainly doesn’t refute it.) By making an official port of your custom software, you can entice others into using your preferred tools. Write a cozy package description to lure other people with similar problems into trying your solution. A few sysadmins will respond with “improvements,” which you should gleefully accept—not because they impact your solution, but because—and this is the important bit—because it means they will have touched the official package. They catch the software’s cooties. You pull them into the damp pit with you. Together you can build better beetle traps and stave off the inevitable for a few more days. Maybe even months.

Making a port is not hard. I’ve done it. I needed a Radius authentication module for Apache, because the alternative was to integrate everything into Active Directory and that would have been even more custom. Not happening. The folks who maintain the ports collection have provided all kinds of instructions in the hope that others will touch it and join them in their much larger, better-appointed damp pit.

If your management still troubles you after all this work, try hissing like a possum.

Have a question for Michael?
Send it to letters@freebsdjournal.org