Notices by Chris Hanson (eschaton@mastodon.social)

@whitequark Nah, mailing lists and newsgroups have a pretty good track record over the past several decades when it comes to running large technical projects. Web forums not so much.

@whitequark I don’t think mailing lists and newsgroups are good for code review either, that’s why PRs were so revolutionary. But mailing lists and newsgroups are vastly superior to web forums for discussion and always have been.

@whitequark Mailing lists and newsgroups are a great experience though, it’s web forums are an absolutely atrocious experience to be involved in. Working on a compiler or code generator should not be gatekept to only those who are willing to put up with web forums.

In all seriousness, the whole “mailing lists and newsgroups are old fashioned, everything has to be a web forum to accommodate people who want to visit a web page to read things” discourse is fucking moronic. If you can’t handle subscribing to or interacting with a newsgroup or mailing list, why should you be trusted to handle the internals of a compiler or code generator? If you really insist on visiting a web page to read the list, use a web-based MUA.

Gods damn but do I hate the “Discourse bot” gamification encouragement bullshit. I had to sign up for the LLVM Discourse forum because they fucking got rid of the mailing lists that had served perfectly well for years and it’s giving me pats on the back for “sticking around and reading things.”

I was on the mailing lists for well over a decade, you stupid machine, and unlike some people I can actually handle reading messages distributed as text without something holding my hand.

Has systemd been rewritten in Rust yet? Seems like that’d be the right thing to do.

@mcc @david42 One of the places where I think Lisp Machines lose is that despite all having OOP facilities as well as decent compilers, they still tended to huge flexible functions rather than super-aggressive decomposition.

@mcc @david42 Have you played with classic Smalltalk-80 or early Squeak? That was from back in the day when the average Smalltalk method length was something like 5-6 lines, and a lot of people felt the system was very tractable like that.

@jwz @dangillmor Let’s hope the Biden administration can get a pandemic response in place that the Trump administration will take credit for.

One of the very, very few things I will not fault Trump for is being proud of “Project Warp Speed.” The insanely rapid development of the various COVID-19 vaccines was a legit moon-landing level achievement by our species. Everyone involved should be proud of it to the point where anti-vaxxers should inspire homicidal rage.

Folks who attended CMU in the late 1980s to early 1990s: Do any of you have still have one of the Andrew environment CD-ROMs that you could get to run Mach and Andrew on personal hardware in your dorm or office?

I know MacMach CDs existed, I gather there were CDs (or tapes) for DECstation, Sun-3, Sun-4, and IBM RT as well. Or did everyone with a workstation bootstrap from the network?

“The “Promise” of “Easier” Programming,” wherein I talk about why 4GLs were temporarily successful, why that was temporary, and why “programming via LLMs” will have the same fate: https://eschatologist.net/blog/?p=374

@thomasfuchs All of that was *a lot* more maintainbale and discoverable than modern rats’ nests of LabView diagrams; LabView mostly won that market on the cost and ubiquity of PC hardware while still interoperating with existing GPIB deployments, not by being better. (And Rocky Mountain BASIC was a solid structured BASIC, not a toy BASIC like on microcomputers.)

@thomasfuchs An enormous amount of factory automation and instrumentation was created in the 1980s using “Rocky Mountain BASIC” on HP 9000-series GPIB instrumentation controllers. (This is what HP’s 68000-based HP 9000-200/300/400 systems were the fanciest versions of.

@thomasfuchs This is one of the things that’s also doomed environments like Smalltalk; plain text has some very, *very* important benefits when used for programming, particularly in the form of long-term maintainability, interoperability, portability, and interoperability. Maintaining one or more “environment” systems over time is much more difficult, simply because comparing between variants is incredibly complex and often not possible except within the system itself.

@thomasfuchs Except—you can! There were large numbers of complex business applications built in Prograph and Sirius Developer, and lots of instrumentation is still built around LabView.

@thomasfuchs The problem with “graphical programming will save us all” is that it tried to optimize the wrong problem: Textual programming complexity. Yes, people have varying levels of difficulty when it comes to using text to program, and some of that is caused by needless complexity in language and environment design. However, a complex application is still a complex application regardless of whether it’s written with Prograph or Swift.

@thomasfuchs There was a cylinder too!