MangoCats

Yeah, I’m in a bigger company, so I have the conversation with the local group, then the various other “Subject Matter Expert” groups that touch the same project in their own way, and just when I think I’m done having it there’s some group that submarines up and says “but we haven’t qualified our code to run on 8 so if you’re going to use it you’ll have to pay us to do the qualification, but we’re really busy so that’s going to be about 6 months before we can get to it…”

adding a preprocessor directive

Yeah, I’ve had to do that a few times lately. Really tweaked me when gcc transitioned things from warnings to errors.

Migrating from 4.6 to .NET Core.

Are we on .NET 7? No, only 8 is available here without unreasonable effort to backdate. Does the .NET 7 code run in 8 without mods? Yes, but that doesn’t change the hassle of talking endlessly about it.

Yeah, 0.5% was what I was offered as Director of 50% of new startups… 0.005% is more realistic - and could range from 0 to outrageous.

More reasonable: roll back to before the datacenter was permitted for construction, find substantially equivalent houses in the general area - 20% more square footage, 20% higher value per square foot, and track their values over time until the present, and whatever the peak was, offer 20% more than that plus moving expenses… effectively 175% of “fair” value + maybe $20K.

There’s the thing: what’s a woodpecker’s property rights on his dead tree? I mean, you can offer him any amount of money and he isn’t likely to just give up the nest he carved in it and move on to try to find another.

Same for people’s homes, there’s a price they could get for them if they tried to sell on the open market because they want to leave, but there’s a very different and often much higher price that would be required to compensate them for the downsides of moving.

When a corporation wants to entice a new hire to move to a new city, they will buy the home at high-end market value, pay the real-estate commissions on both sale of the old home and purchase of the new home, pay for professional movers, and bulk up all of these benefits to cover increased income taxes where applicable. And that’s for a new hire who wants to move.

If China manages to scale their hardware production, this could finally be “their day in the sun” where they clearly surpass the West in the way the West has outshone them for 100 years. MoE has its place, but a MoDE where each E is itself a dense model would be more powerful still, mostly you need the silicon gates and power to drive them.

Three Gorges makes hydro-power, right? 22MW -> 100 TWh per year, just from that one structure, I bet that will run at least one AGI… fredgao.com/…/deepseeks-liang-wenfeng-breaks-his

Back in the day, I was getting a new “99% compatible” DOS version every 6 months (until we transitioned to the PharLap 32 bit extender, got more memory to work with AND stopped the perpetual upgrade treadmill.)

That “99% compatible” thing means: in a 10,000 LOC codebase, you’ve got 100 things to fix before it works on the new DOS.

I really liked working with Qt from 2006-2024 or so, the only major break was 4-5 (must admit, I never migrated to 6, and nobody forced me to…) and the migration from 4-5 was less painful than a DOS version bump. Also, Qt fully insulated our code from garbage changes happening at lower layers.

It saves you from having to type

Ooooh my poor fingers… yes, I used to worry about abbreviating everything as short as possible, and I still control my loops with i j and k, but when it comes to ambiguous var vs QString or int32 or float or double or whatever type du jour is, I’ll take the time and effort to write it out rather than taking the time later to untangle how the ambiguity might have resolved and all the unexpected things that might happen as a result of an unexpected resolution.

I have definitely seen MS Office and Visual Studio stall at a level and stay there until after they’re clearly surpassed by open source competition, then they fund another round of development and … change, some might say improve, but often it looks like change for the sake of change, especially in Office.

particularly in Europe, where there is a growing desire to break free from the US SaaS stranglehold, using open source software.

I’m sorry, but it took EXTREME provocation to get Europe to start to talk about moving, again. Previous starts to move to open source software across Europe have mostly faltered and reverted to the comfort of handing cash to foreign companies.

racing toward free once it becomes good enough.

I don’t know… I was using Open(now Libre)Office over 20 years ago, it wasn’t just good enough, it was better than MS Office at the time, and yet… even though LibreOffice has been free and far more than “good enough” for long over a decade, my corporate decision makers insist that we all use Office365 subscriptions, complete with their service outages and other issues.

I see a business model where “we’re done, this one is (finally) good enough and now we’ll stop bleeding cash on the training and turn up the screws on the customers we’ve hooked on loss leader pricing.” Open weight models will never stop training for improvement, the costs for training seem to be inexorably falling, and any business model built on the idea that they can kick back and roll in the profits after their initial “hard work” is going to lose all their customers to better products.

This isn’t some captive market like US automobile customers who have no choice but the limited selection of crap that is put in front of them. At least not as long as the internet remains relatively open.

My point is: as a programmer, you should always care how your value is handled. All variable types have limitations, edge cases, and resource demands (infinite capacity big-int representations are slow and memory intense, even if they never overflow or underflow.) KNOW what your code is doing with the value.

I’m O.K. with API style functions accepting all kinds of inputs and dealing with them in a predictable way once called, but if you’re writing code that handles values, once you have been assigned those values you should be getting, deterministically, the same result every time.

Otherwise, you’re like those kids in programming class writing while loops with floating point control variables that increment +0.1 per loop, thresholded to stop at v >= 100.0

My perception is one of a treadmill. My fellow developers in C# are much more frequently analyzing migration from one incarnation to the next, updating and installing their license files, evaluating compatibility between versions, and generally spinning their wheels on things not required for getting work done on other platforms.

I come from a background of porting code between architectures, like 16 to 32 bit. The int type in C was (is) ambiguous in that case, 16 bit signed int on the 16 bit system, 32 bits on the 32 bit system. First thing I had to do to make the 500,000 LOC package start to run properly when ported from 16 bits to 32 bits was find and replace all " int "with " int16 ", “(int)” with “(int16)” etc. It was only going wrong “quite rarely” - but “quite rarely” doesn’t cut it, they ALL needed to behave predictably the same as they did before the move. I believe that effort took 3 solid work days, 24 hours of my life, because “int” was “good enough” for the previous team.

People share sensitive information all the time in places and ways they shouldn’t - they think noone is looking, noone is listening, noone is recording - the problem is: when it goes on the web, it’s recorded, replayable, visible, for all to see.

It’s a tough balance, it puts cops in a position to destroy someone’s life with zero proof of any law being broken - like Paul Reubens…

The fact that he’s in the news is a VERY good sign. In a real 3rd world shithole he would just disappear because he pissed off an unaccountable field goon.

I tried to use C++ in 1991 but the Borland Turbo C++ compiler was too buggy, so we used C instead. By 1996 C++ was “ready for primetime” on IBM-compatible PCs (yes, they still called them that, sometimes in 1996) and I switched - for 30 years.

I just spent 10 days building a proof of concept in Python, all the “common wisdom” says that was the fast way to get it done. Now that the prototype is done and the (initial) user feedback is addressed, I’m running a port to something better for lightweight, performant, easy deployment… sadly, C++ isn’t even on the radar for potential targets - top 3 candidates were Go, Rust and C#. I can’t abide the C# ecosystem, and Rust is just a little too rigid and immature for my tastes, so here we Go… anticipated to take 50% longer to port from Python than the Python took to develop in the first place… we shall see…

Typeless variables are a delusion, they’re just guessing what type they should be for you.