I run Linux as a desktop and have done so since around 2008. I started with Ubuntu, and after a while (probably around 5-10 years) I moved to Debian. Every once in a while, I'll read about one of the new Arch-based distros (Manjaro, Calyx OS) and decide to give it a try. After about a few hours, I realize that some of the programs I use on a regular basis are not available (easily) outside of the AUR. When you read about the AUR as an intermediate user, you understand how dangerous it can be, but you feel like it's necessary to use Linux as your main computing device. There are applications that are packaged as DEB/RPM but not for Arch, and are not available as Flatpaks (or AppImages or Snaps). Some of these are proprietary.

One in particular which comes to mind is Insync, which I use to synchronize Google shared folders to my home directory. It is much easier to use than rclone and the latency is a lot lower. If I move to an Arch-based system, I have to get that from the AUR. Now, I do feel like I have the experience to read the PKGBUILD and audit it for weird stuff going on, but I'm also not arrogant enough to believe that someone could not sneak something by me.

I use Debian Stable, and all of my software is available. Some of the software is dated, obviously; I'm running KDE 6.3.6 and kernel 6.12. But in general, I don't have huge issues with that, and if there was an application I needed to update, I probably could do it either with Flatpaks or compiling from source. Honestly though, I cannot remember the last time I needed to do that. Maybe it helps that I'm not a professional software developer and I don't need access to the latest versions of everything. I also know that some Debian users address those issues by running testing or unstable.

There's a part of me that wonders if these attacks are related to the surge in popularity of Calyx OS. I teach high school, and I noticed last year that one of my ninth graders was running KDE on his laptop. I asked him what distro he was running, and he said Calyx OS. I was surprised by that - most of the time when I run into a high school kid they're running something in the Debian family (including Ubuntu and its derivatives).

I am not sure it is fair to describe this as no better than a Chromebook. This machine is running macOS. You can put Homebrew on it, and install many applications that way (including many open source applications). I use Geany, Octave, Maxima, R, and a whole other host of applications on my Mac, which I use mostly at work. Most of the applications I use on my Debian desktop are available on Homebrew. If you're concerned about Homebrew for security reasons, you can usually install packages directly from the application website.

I totally understand people saying they don't want to run macOS because of various reasons, but the "walled garden" description of macOS is not fair in my opinion (it is completely fair for iPhones and iPads). I'm able to install the software I want on my Mac at work. While the security settings of macOS make you jump through some additional hoops, I don't think it's an overwhelming burden.

I have heard this machine described as being on par with an M1 Macbook Air. My wife uses that machine on a daily basis, and it works well. My daughter is using an M1 Pro machine, and does not want to upgrade for college because she feels it is unnecessary.

I think we need to do a better job of going after Mac for the real issues it has: cost of upgrades, lack of repairability, and inability to install Linux on the machine.

Math teachers (I teach high school calculus and computer science) have struggled with these issues for about ten years. Tools exist that allow you to photograph algebra problems and generate not only the correct answer, but all of the supporting work. When it comes time for the test, these students do not usually do well, unless they're smart enough to learn the material only through the in-class examples. However, I've found that by the time you get to Calculus II, that stops working even for the brightest of students. Most of the high-level students eventually figure this out. When I teach Calculus III (multivariable) at the high school level, I intentionally don't check homework at all (I'll provide assistance, obviously), so that students learn the consequences of not doing it, and develop intrinsic motivation to practice.

The use of generative AI is probably going to change college assessments. There will be more emphasis on timed, in-person performance assessments in all subjects, similar to what happens in technical interviews for engineers and computer scientists. The sad part of this is that there many individuals who do not perform well in high-pressure environments like that, and they will find it increasingly difficult to succeed in college courses. I remember taking upper-division math courses where we would be given a week to do a take-home final that was essentially a series of proofs. Those days may be coming to an end, and it's very unfortunate because that was a much more authentic, deeper mathematical practice than a timed test. I recall taking walks around campus, thinking about the problems, and often it would take a mile or two before a path to a proof would present itself.

Not all institutions have the same initial programming pathway, but it seems like a lot of US institutions (according to a CS professor friend of mine) do three foundational courses: CS 0 (Introduction to Programming), CS 1 (Object-Oriented Programming), and CS 2 (Data Structures). CS 0 is often a general education course designed for non-majors that majors with HS experience skip over. That one can be taught in any language and probably Python makes the most sense. I do think it is helpful if CS 1 and CS 2 are taught in the same language. I warn my AP CS A students (who learn Java) that if CS 2 is being taught in C++ at their university, they should probably retake CS 1 even if they already have AP credit for it. At least one of my former students who just finished undergrad took CS 1 in Java and CS 2 in C++ at the same institution (he said it made everything about the course harder).

For the parents of CS students out there, I do think AP CS A prepares students well for Data Structures. I've probably a dozen students who passed the AP CS examination as juniors to a local university for Data Structures during their senior year, and they typically get As and Bs. Sometimes they've been the best student in their sections, probably because AP CS is taught in a full year so the students get a lot more programming experience; we have time for more projects.

I agree with you -- I feel that the challenges have been getting harder. I've done various of these going back to 2015. In 2020, I actually did the whole thing (one of my solutions was pretty inefficient, but it worked). This year, the fourth day, second part requires some skills that first year college students don't have (skills taught in a data structures course). I teach AP CS A, which is generally the course right before data structures. I won't say more because some people here probably haven't done it yet but intend to do it later.

The person you responded to mentioned using regular expressions. I'm embarrassed to admit that I never learned regular expressions (I understand the idea, but never learned the practicalities of using them). It just never came up. So I actually do everything for these using the methods of the String class. That sometimes is a big hassle, and often (including today) I say to myself, "you should take the time to learn how to use regular expressions."

I'm not sure I understand why they are worried about cheating -- people often post their solutions on GitHub or alternatives. It's just for fun. If I get frustrated, I stop. Sometimes after thinking for a while about it, I come up with an alternative approach that works better.

I teach this course and this year for the first time I was a grader. I graded Question 1. The week after I graded, I attended a professional development session led by a professor who has done the reading for around 30 years (back into the days of Pascal). For those of you not familiar with AP, the readers are typically divided about 60/40 between high school teachers and university professors. The exam committee is similarly split (although that one I think they keep at 50/50). The professor I was with addressed many of the points above which are valid. I think they may have had an "off" year. CS professors and teachers are just as opinionated about some of these things :)

Several posters above made good comments about the detailed requirements specification. I agree with their points, and the AP CS people are also aware of it. The amount of text present per question is being discussed on a national level. We practice AP questions like this in my class so that the students are familiar with the exam format. I tell my students that careful reading is a necessity. But in this case, 22.5 minutes per question seems short to me.

The questions themselves were hard this year. Each question is graded on a nine-point rubric so partial credit is a definite thing (some of those nine points are fairly easy to get). Those rubrics and student samples get released in September to the public if you are curious (just search 2023 AP Computer Science A FRQ Scoring Guidelines).

Another poster made the comment about there being two correct ways to finish a problem -- and I agree, that was dumb, and it is unusual. I've looked at exams going back to 2009, and almost always you are told "this is exactly the state it should be in when finished."

They are also struggling with whether the exam format should be revised. They piloted a project/portfolio exam this year but then the whole ChatGPT thing happened. And they seem terrified of digital exams (probably because most schools don't have dedicated, controlled, computer labs any more). They are convinced that some students would find a way to cheat (esp. in a course like AP CS A). Some students who have issues writing by hand are allowed to type their exams, but it's a relatively small number.

The 30+ years refers to the total of all of the sentences. It seems as if they decided to start their own streaming service using pirated content. A couple of the guys made seven figures (in pounds, not dollars) on this. This is not one guy stealing cable in his house and watching the big game (or do you say "match" in the UK?).

I understand that the official provider was probably ridiculously overpriced, but I'm not sure how you could think you'd get away with this. They must have had some serious infrastructure running this.

I used to own a Surface Pro 6 at work. I tried putting Linux on it using the work of the linux-surface project at GitHub. They have had a heck of a time getting camera support working under Linux on many of the models. And they've spent years trying. I'm not sure what's so special about the cameras in surface devices -- I suspect it has something to do with them being capable of handling the facial recognition Windows Hello uses. You can go to this page to see what I mean. Only the first three generations of surface tablets have cameras working under Linux.

If I record a 45 minute lecture on my Mac using the built-in recording application (I have a M1 air) I get a video in excess of 1 GB. ffmpeg reduces that by an order of magnitude, with a very simple terminal command. And I'm not an expert on video editing or using ffmpeg at all -- I bet others could do better.

I cannot get over the decision Ubuntu made to go with Snaps. It's the main reason I run Mint. Mint has its issues also, but I cannot abide the idea that Firefox cannot start quickly on an Ryzen 5850U. On Mint, the dosbox package has some issues, so I grabbed the dosbox-staging flatpak. It starts immediately, no issues -- completely different than the Firefox snap experience. If someone here wants to explain to me why Canonical has invested so much in Snap, I would love to here the explanation. Maybe there is something I'm missing, but I've never had issues with either Appimage or Flatpak.

I have been following Asahi Linux because I also have an M1 mac. I have a separate machine running Linux, so I'm fine with staying with macOS. I remember watching a long question/answer session on YouTube where Linus is at a Debian conference, and he says that he is "not an IT guy". The point he seemed to be making at the time is while he obviously enjoys working on the development of the kernel, he doesn't have time to waste debugging the other small, common issues that arise with most Linux distributions (he was complaining about root access being needed to add a printer on one of his kids' machines, IIRC). Asahi is apparently in good enough shape that he's willing to use it. I know they don't have the GPU figured out yet (they seem to be making progress) but from what Linus has described, his work environment is a bunch of terminals, so he's probably fine with that.

One of the things that concerns me about ARM and Linux is that a lot of the software that exists for Linux and is (usually because it is nonfree) not included in package managers is only available for x86. Wine is also an issue (although I did manage, on a Raspberry Pi 4, to get Box86 + Wine working together well enough to run some simple Windows applications). Even when Asahi gets to the point of fully supporting the M1 hardware, that's going to be an issue for people who are trying to just run Linux and do everything they need to do. I have a license for Microsoft Office 2007 (which is good enough for most Office documents) that works flawlessly in Wine, but it's not going to work well on a Linux machine running ARM (although maybe the M1 would be powerful enough with Box86, who knows).

There's been a lot of discussion on the economics of installing rooftop solar. I thought I would give my perspective because we installed solar on our house about a year ago. Our system cost $30,500. It was financed for 1.5% APY for twenty years with zero down. The monthly payment is about $150. We received a tax refund of $8322, which we did not apply to the principal, since the interest rate is so low. You should know that if you redo your roof at the same time, you can claim the part of the roof that is underneath the panels. That's why our credit is a little higher than 26% of the system cost. Our cost was higher, because we do not have a south-facing roof. Our panels face to the west. We live in Wisconsin, which is not the best environment for solar power either.

So, in terms of economic impact to us, we started with a "balance" of 8322. Our electric bill has been reduced in two ways. One, power we generate and consume is subtracted from our bill at the rate the utility charges. Two, power we generate and sell back is subtracted from our bill at 1/3 of the wholesale rate (that varies from state to state). When you take these amounts for the last year, and subtract our loan payments, the net expense is about $1100 per year. I have spreadsheets tracking this because I am curious about the economics also; we did it more for environmental concerns.

This means it will be about 7.5 years before we actually "pay" anything for our solar installation (before we wipe out the $8322 in cash we were handed for installing the system). After that 7.5 year grace period, we spend $1100 per year helping the environment. Assuming we stay in our house for the full length of the loan, we would then be saving money from that point onward.

There are some other economic factors which probably reduce that cost. First of all, the $8322 is being invested in a conservative portfolio which typically earns a return around 6%. Second, we think (we're not sure) that our house is cooler in the summer since we installed the panels; it definitely is not making the house hotter. Third, the forecasting is conservative because it assumes the cost of electricity does not increase (unlikely). Fourth, while I have a hybrid vehicle now, if we get an electric vehicle, we will save more money, because we'll be using more of the electricity we generate instead of selling it back.

Finally, we give a considerable amount of money to charity, and we view this expense as a component of our charitable contributions. That's probably the most important point I would make here. The other point is that installing this system does not make sense if you anticipate relocating to another area. If we did move, we'd probably realize some savings also from the increased value of the house, but it's hard to say.

I have a bachelor's degree in electrical engineering, and a master's in computing (that was an interdisciplinary program which in my case focused mostly on graduate work in analytics and databases). Generally in education in the United States, the presumption is that teachers in a discipline are one-level-above their students. High school teachers typically have an undergraduate degree in the subject they teach; sometimes, if a teacher has multiple licenses, one of they may only have a minor in the subject. Undergraduate instructors have a master's degree in the subject they teach; and (ideally) graduate instructors have a PhD (this one is often violated, I've had lots of graduate courses taught by people with just master's degrees).

Speaking as a department head, it's hard to find people with undergraduate degrees in CS who are willing to teach. That's why there's so much non-degree professional development aimed at creating CS teachers (many of these programs try to recruit math teachers because math teachers are generally analytical thinkers and many have had one programming course). Some of this PD is good, some of it is too dumbed-down, in my opinion.

Based on the limited availability of candidates, my criteria for someone who wants to teach high school computer science is that they should at least have a course in object-oriented programming and a course in data structures. If they want to teach something above AP CS A (some schools teach data structures to their students as a follow-up to AP CS A) they should have an undergraduate degree in CS. This ensures that intelligent students who ask deeper questions about the material can be served appropriately. Last week I taught recursion to my students, and I was able to talk about some topics (like memoization) that many books omit. Having been to many training sessions with AP CS A teachers, I doubt even the ones who have a lot of industry experience (which doesn't mean they have a degree in CS, incidentally) can give a good explanation of those topics.

More importantly, though, the teacher should like to write code; he or she should be passionate about the subject. I do a lot of recreational programming, like Advent of Code and Project Euler (I'm also a math teacher), and I enjoy solving problems.

I have been thinking about getting a new laptop with a Zen 3 processor -- do the most recent versions of the Linux kernel support the "preferred core" feature?

I am no big Microsoft fan (I run desktop Linux at home) but I gave this a try because I teach two courses using Java at the high school level. This offering has a few advantages:

1) It automatically adds itself to the system path in Windows, which means the students just need to install this and a lightweight IDE like Geany and they are ready to compile Hello, World*.

2) It does not require users to create an account to download the installer (unlike both the Oracle JDK and the Red Hat OpenJDK I had been using because Red Hat also has their installation put java/javac on the system path).

3) It allows me to use the same product on both macOS and Windows (and the same IDE also, since I use Geany typically). Red Hat's OpenJDK does not have a macOS installer.

4) It is licensed GPL2 so I don't have to worry about distributing something with bizarre license restrictions (like the Oracle JDK).

This will no doubt make the first day of future semesters a lot easier to manage. I would never use this myself for my own work because I just grab OpenJDK from my distribution's repos, but it will simplify things for instructors of elementary programming classes.

*Of course, we still have to edit the system path if the student has installed the Oracle JDK because they are on our FIRST Robotics team, which will conveniently add "java" but not "javac" to the system path. I wish Oracle would stop doing that. Either set up the paths correctly or not at all.