The problem with microcontroller development...

[westfw]

When I was a 1st year university student

Heh.  I handed in at least two major "final projects" written entirely in PDP10 assembly language.It wasn't until MUCH later that I noticed that this might have been a really mean thing to do to the profs/graders (who may or may not have had any knowledge of PDP10s.)
Aced the classes, though!

[PlainName]

HDDs no longer have 70MB, my gawd.

You need to at least double, preferably triple, whatever size you end up with for backups.

So a compact, easily manageable, 5-minute setup, complete ecosystem, is a problem for you.

"Compact"? Pretty much all the manufacturer IDEs are bloaty things which assume the end user will be using it in the same way the developer does (or, seemingly, thinks he would if he actually used it). The editors are complete shite, nothing is consistent, stuff is hard to track down when it goes wrong and they shit all over your system drive.

And then just as you finally manage to put up with it all, they change the damn thing.

Thing is, for the chip manufacturer the development tool isn't a core product. And it shows.

[SiliconWizard]

Thing is, for the chip manufacturer the development tool isn't a core product. And it shows.

Well, it's a key marketing thing, but sure not the core of their business. Not sure the investment/effort is the problem here. The problem, IMO, is just that's it's big software, and big software sucks these days, wherever it comes from, pretty much.

I personallly think vendors should focus on providing good, correct and *exhaustive* support files for their chips (memory map, complete register definitions including all bits, linker scripts, startup code, etc), clearly document the compiler options required (march/mabi/etc), and that's it. And a few clear example projects using make or Cmake. They may additionally provide extensions for VSCode or whatever, saving time for people using that. And that would be it. No need to maintain their own "IDE". Just a thought. It's a waste of their time, and, in the end, that of developers (when they end up realizing how much time they have wasted).

[DavidAlfa]

You need to at least double, preferably triple, whatever size you end up with for backups.

That's not the way you make backups, 3 copies of the same thing in the same drive, makes "pop" and all copies are gone.
Ever heard of git or svn? Or just Google Drive.

Anyways, space this is irrelevant when you can buy 2TB drivers for peanuts.

"Compact"? Pretty much all the manufacturer IDEs are bloaty things which assume the end user will be using it in the same way the developer does (or, seemingly, thinks he would if he actually used it). The editors are complete shite, nothing is consistent, stuff is hard to track down when it goes wrong and they shit all over your system drive.

And then just as you finally manage to put up with it all, they change the damn thing.

Not really. Everything is well contained i.e. C:\ST, or Program files\Microchip.
IDE editors are great with syntax and real time checks, Code assist, go to declaration/definition...
There's no such  "as you finally manage to put up with it all, they change the damn thing.", setup takes 5-10 minutes and you're done.
Maybe every 10-15 years something changes, that's progress. Otherwise we'd keep writing hex machine code manually.

ST is a bit of an exception here as they had 5 IDEs in 10 years before finally staying with CubeEIDE.
Still, the libs were the same, so it wasn't a terrible drama.

Sounds more like "Let me be, I don't wanna learn anything new" thing.

[PlainName]

You need to at least double, preferably triple, whatever size you end up with for backups.

That's not the way you make backups, 3 copies of the same thing in the same drive, makes "pop" and all copies are gone.
Ever heard of git or svn? Or just Google Drive.

Ah, someone that doesn't do backups. And thinks a reinstall of everything is a 5 min job.

You would be using removable drives, and since you would never write a backup to the last media used you would need at least two. Plus something like a NAS for on-the-fly and/or automated jobs (you aren't going to remember or bother to switch removeable drive every day or whatever). So that's minimally three drives you've got to upgrade for the new 'not very big really' requirements.

[brucehoult]

Backups are for wimps. Real men upload their data to an FTP site and have everyone else mirror it.

[mikeselectricstuff]

Inevitable I'm afraid - vendors have to support their tools, so need to have a well-defined configuration, even though this will usually have lots of characteristics inherited from earlier versions, possibly going back decades.

Of course users are free to adapt & do what they want, but from experience  this becomes problematic long-term when support for new parts is needed, or projects need to be useable by others.

This is a particular issue where an external consultant/contractor delivers code to a client who wants to be able to maintain it.  "Just install manufacturer tool version x.yy and load the project" is all they want to hear.
Say what you want about Microchips tools, but IME the above "just works"  for everything from a 6-pin SOT23 to a 32 bit 200MHz device.

[tszaboo]

So a compact, easily manageable, 5-minute setup, complete ecosystem, is a problem for you.
Yeah, better to spend 5 hours setting everything up manually, crossing fingers nothing screws up someday.

My IDE is broken! Reinstall, back to work in the time I took a coffee.
Terrible.

Seems to me you'd keep climbing the mountain with an old rope to reach home everyday instead using the new electric elevator.

Why is this matter coming up so often?
HDDs no longer have 70MB, my gawd.

This goes for also the others that say: "Why don't you do X?"
Most Firmware developers I've seen work in teams. Teams have rules. You might not have any say in what tool to use to program your micro.
You might be working on legacy projects that you are not supposed to change much, just "Make it work with extra X" that are decades old. Developed by engineers that didn't care, don't work here anymore.

If you don't work like that, we are happy for you.

[mtwieg]

So a compact, easily manageable, 5-minute setup, complete ecosystem, is a problem for you.
Yeah, better to spend 5 hours setting everything up manually, crossing fingers nothing screws up someday.

Yeah I suspect that even if I found a way to build without installing CCS, they would immediately see it's an even worse option and reject it. Not sure what they're actually hoping to get.

Inevitable I'm afraid - vendors have to support their tools, so need to have a well-defined configuration

Yeah to me this is the true benefit of a vendor-provided IDE. It packages many things (compiler, linker, debugger, libraries, maybe an SDK) in a tested configuration. Editing code in the IDE is basically its least important function (just use VScode for that).