Even with the MAX and the recent (last ~2 years) spate of incidents, flying is safer now than it ever has been, and certainly safer than it has been over its lifetime.

A kindle is more like a bookshelf from IKEA, it has a finite lifetime.

For (others) reference the oldest bits of code in our software are from the mid-late 90’s and the oldest systems still paying for support rely on parts to build that is not available at any price, its all just made of unobtainium, whereas I can sell them a brand news that does everything they have today and more.

Some of those customers cannot be upgraded without hardware replacement, we can sell them a brand new system that will do everything (and more) their old one will, but they dont want to spend the money, and we are happy to take the money for support (the old CAPEX vs OPEX argument).

Some of this is sorta easy, its COTS hardware, but we also have much older systems that due to component obsolescence I simply cannot build replacements anymore. 10+ years of support ought to be enough for 90% of the products out there, at some point the answer really is “upgrade your hardware” - we didn’t sign up to indefinitely support not just hardware (much of which we cannot build spares for) much less the software ecosystem around the hardware.

Long term I plan on increasing support renewal costs for systems that are older than 10 years old to encourage hardware refreshes.

Like I still have to have XP VM’s to build firmware for older products, when is it reasonable to cut off service and support?

The function of if tobacco causes cancer has as much to do with processing (it used to be cured by wood fire at a higher temperature which is where much of the carcinogenic properties came from) and the byproducts that processing creates, particularly Nitrosamine, its now cured differently in a process which is closer to snus, and somewhat safer.

Nicotine addiction (which I have) should be about harm reduction first, cigarettes are the only product that I can think of if used as commonly used will kill you or dramatically shorten your life, and it probably wont be cancer, it’ll be COPD, heart disease, or other cardiovascular issues - which are the same issues firefighters get from repeated smoke exposure. Breathing the byproducts of combustion is what’s really awful (and deadly).

https://en.wikipedia.org/wiki/Nitrosamine

Pretty much every other form of tobacco that is not cigarettes is less bad.

Try estimating doing win11 updates on a 20 year old piece of delphi spftware with hardware full go custom ASICs be expected to lsat?

I'm just in the process of developing a lifecycle policy, being able to cut off support for a 12 year systems would make my life much more full of joy.

Its 'light manufacturing' for a company that makes custom formed acrylics for aerospace.

https://www.google.com/maps/place/33°47'00.8%22N+117°59'59.8...

In my working life I've seen -

at least 3 different ways to do RS-232 on 8p8c

DTE-DCE is always an issue on standard 232, otherwise I wouldn't have so many null modem cables

232 and 3.3v TTL on the same board or assembly

3.3v and 5v TTL on the same board or assembly

Inconsistent labelling.

I'm in my mid 40's and I think there is a reasonable chance something with RS-232 serial timings will outlast me, it'd be nice to make it more foolproof, as its one of a very few interfaces that will work without drivers.

I do think as a matter of standard good design practice we should be putting clamping diodes on debug ports to prevent blowing things up if hit with the wrong voltage,

1. Automatic DCE-DTE detection and an interface which will rewire itself as needed to be the correct way, or you automatically know DCE vs DTE by connector gender.

2. Automatic Voltage Detection - 232 levels, TTL 5v, TTL 3v - and interfaces that are isolated enough to deal with the wrong voltage (clamping diodes or whatever), or different cable sizes for each.

3. Automatic type detection - TTL/RS-232, RS-422, RS-485, or different connector types by each.

Ideally I'd do this on a 8p8c or 10p10c connector, because of ease of making cables, with various resistance values across pins 1-8, or 1-10 to tell you what kind of interface it was.