{"@attributes":{"version":"2.0"},"channel":{"title":"Debian on Website of Jamie McClelland","link":"https:\/\/current.workingdirectory.net\/tags\/debian\/","description":"Recent content in Debian on Website of Jamie McClelland","generator":"Hugo -- gohugo.io","language":"en","lastBuildDate":"Fri, 10 Apr 2026 08:27:10 -0400","item":[{"title":"AI Hacking the Planet","link":"https:\/\/current.workingdirectory.net\/posts\/2026\/ai-hacking-the-planet\/","pubDate":"Fri, 10 Apr 2026 08:27:10 -0400","guid":"https:\/\/current.workingdirectory.net\/posts\/2026\/ai-hacking-the-planet\/","description":"

A colleague asked me if we should move all our money to our pillow cases after\nreading the latest AI editorial from Thomas\nFriedman<\/a>.\nThe article reads like a press release from Anthropic, repeating the claim that\ntheir latest AI model is so good at finding software vulnerabilities that it is\na danger to the world.<\/p>\n

I think I now know what it’s like to be a doctor who is forced to watch Gray’s\nAnatomy.<\/p>\n

By now every journalist should be able to recognize the AI publicity playbook:<\/p>\n

Step 1:<\/strong> Start with a wildly unsubstantiated claim about how dangerous your\nproduct is:<\/p>\n

AI will cause human extinction before we have a chance to colonize mars<\/em><\/del>\n(remember that one? Even Kim Stanley Robinson, author of perhaps the most\ncompelling science fiction on colonizing mars calls bull\nshit<\/a>\non it).<\/p>\n

AI will eliminate all of our jobs<\/em><\/del> (this one was extremely effective at\nproviding cover for software companies laying off staff but it has quickly\ndawned on people that the companies that did this are living in chaos not\nhumming along happily with functional robots)<\/p>\n

AI will discover massive software vulnerabilities allowing bad actors to “hack\npretty much every major software system in the world”.<\/em> (Did Friedman pull that\ndirectly from Anthropic’s press release or was that his contribution?)<\/p>\n

Step 2:<\/strong> To help stave off human collapse, only release the new version to a\nvetted group of software companies and developers, preferably ones with big\nsocial media followings<\/p>\n

Step 3:<\/strong> Wait for the limited release developers to spew unbridled\nenthusiasm and shocking examples that seem to suggest this new AI produce is\ntruly unbelievable<\/p>\n

Step 4:<\/strong> Watch stock prices and valuations soar<\/p>\n

Step 5:<\/strong> Release to the world, and experience a steady stream of mockery as\npeople discover how wrong you are<\/p>\n

Step 6:<\/strong> Start over<\/p>\n

Even if Friedman missed the text book example of the playbook, I have to ask:\nif you think bad actors compromising software resulting in massive loss of\nprivate data, major outages and wasted resources needs to be reported on, then\nwhere have you been for the last 10 years? This literally happens on a daily\nbasis<\/a> due to the\nfundamentally flawed way capitalism has been writing software even before the\ninvention of AI. A small part of me wonders - maybe AI writing software is not\nso bad, because how could it be any worse than it is now?<\/p>\n

Also, let’s keep in mind that AI’s super ability at finding vulnerable software\ndepends on having access to the software’s source code, which most companies\nkeep locked up tight. That means the owners of the software can use AI to find\nvulnerabilities and fix them but bad actors can’t.<\/p>\n

Oh, but wait, what if a company is so incompetent that they accidentally\nrelease their proprietary software to the\nInternet<\/a>?<\/p>\n

Surely that would allow AI bots to discover their vulnerabilities and destroy\nthe company right? I’m not sure if anyone has discovered world ending\nvulnerabilities in Anthropic’s Claude code since it was accidentally released,\nbut it is fun to watch people mock\nsoftware<\/a> that is clearly\nwritten by AI (and spoiler alert, it seems way worse that software written\nnow).<\/p>\n

Well… we probably should all be keeping our money in a pillow case anyway.<\/p>\n"},{"title":"Mailman3 has 2 databases. Whoops.","link":"https:\/\/current.workingdirectory.net\/posts\/2026\/mailman3-migration\/","pubDate":"Mon, 30 Mar 2026 08:27:10 -0400","guid":"https:\/\/current.workingdirectory.net\/posts\/2026\/mailman3-migration\/","description":"

At May First<\/a> we have been carefully planning our\nmigration of about 1200 lists from mailman2 to mailman3 for almost six months\nnow. We did a lot of user communications, had several months of beta testing\nwith a handful of lists ported over, and everything was looking good. So we\nkicked off the migration!<\/p>\n

But, about 15% of the way through I started seeing sqlite lock errors. Wait,\nwhat? I carefully re-configured mailman3 to use postgres, not sqlite. Well,\nyes, but apparently that was for the database managing the email list\nconfiguration, not the database powering the django web app, which,\nincidentally, also includes hundresds of gigabytes of archives. In other words,\nthe one we really<\/em> need in postgres, not sqlite.<\/p>\n

Moving from sqlite to postgres<\/h2>\n

Well that sucks. We immediately stopped the migration to deal with this.<\/p>\n

I noticed that the web is full of useful django instructions on how to migrate\nyour database from one database to antoher. However, if you read the fine\nprint, those convenient looking “dumpdata<\/code> loaddata<\/code>” workflows are designed\nto move the table definitions and a small amount of data. In our case, even\nafter just 15% of our lists moved, our sqlite database was about 30GB.<\/p>\n

I considered some of the hacks to manage memory and try to run this via django,\nbut eventually decided that pgloader<\/a> was a more robust\noption. This option also allowed me to more easily test things out on a copy of\nour sqlite database (made while mailman was turned off). This way I could\nmigrate and re-migrate the sqlite database over and over without impacting our\nlive installation until I was satisfied it was all working.<\/p>\n

My first decision was to opt out of pgloader’s schema creation. I used django’s\nschema creation tool by:<\/p>\n