And never mind getting a nice IDE environment like Visual Studio up and running.
So it may just be a matter of familiarity and/or the particulars of the projects you work on.
Free alternatives would be Visual Studio Code, which is pretty ok even for C++, or KDevelop, which almost kind of works, or SpaceVim / SpacEmacs and about a week of learning how to use it. I'm in the final camp, since basically every serious C/C++ dev on Linux has a Vim or Emacs setup they've built themselves that does about 90% of what Visual Studio gives you out of the box, with some things better and others worse. It can work pretty well, but you have to understand it's not VS and it works in its own way.
If you have the money, you can buy a good IDE for Linux. Otherwise, learning to make your own out of high-level building blocks is also a useful skill, but takes a little bit of time. Personally the first time I had to write C for Linux out of college it did take me about a week before I was up to speed with it (but I was already sort of familiar with vim as a pure editor).
Small parts of Qt for specialized applications are under GPL, and abiding LGPL in embedded hardware is hard, but for normal desktop software it's not a problem.
No, you do not need a commercial license for creator, neither did you at other points in the past few years.
Even though I would argue QtCreator is only bearable when working with Qt.
I guess I'll just have to settle with using Visual Studio Code for linux.
If you're willing to sacrifice the stability and performance of your output for a saccharine IDE and its installer then that really speaks more to your priorities as an engineer.
You mean like
- gdb randomly sucking up 16G of ram - gdb crashing with internal errors - clang/gcc miscompiling my code - perf refusing to record stacks - perf refusing to symbolise stacks because it's having a disagreement with the compiler people about debug data formats.
- locking up X11 and having to SSH in to fix it (because, you know, X11 doesn't have per-application input queues, unlike WindowsNT3.5)
- and a ton of other minor little annoyances with hardware support that Mac/Windows ironed out years ago
I like Linux a lot, the tools are great, but let us not pretend they are all that, they still have a way to go.
I feel like developing on Windows is more difficult because of all it's abstraction layers. WMI, DCOM, the registry, overlapping snap-in functionality, ect.
To me programming on Windows feels like you've got to peel apart all these layers to understand what's between them. Linux is one layer and you can already see right through it.
> - and a ton of other minor little annoyances with hardware support that Mac/Windows ironed out years ago
Uhh, did you just say that Mac has ironed out harware support? For what? They only support their own hardware. The hackintosh community will be the first ones to tell you that Apple purposely doesn't iron out hardware... ANYTHING unless it already says "Apple" on it.
However, Windows has great tools too. And by that, I mean Visual Studio. In particular, I really miss its debugger when I am on Linux.
It is just a different philosophy with its pros and cons. On one side, you have a lot of small, loosely coupled tools, text files and command lines. On the other you have a well integrated monolith and GUI.
There is also something easy about GUI based development and debugging on Windows. A runtime environment that is fairly standard for gaming is also a big plus.
https://en.wikipedia.org/wiki/Development_of_Doom#Programmin...
NeXT was BSD with a Mach kernel.
This claim has mutated so grossly over the years it's comical. I clearly recall reading back in the day, probably in Dr. Dobbs Journal or Game Developer Magazine, that the NeXT Cube was largely used just for tools like the level editor, and they were generally seeking uses for the novel machine.
It was not the primary development environment. They only had one of these expensive machines after all, and the photos I saw from the era were of designer Tom Hall using it [1].
Do you really think John Carmack and Tom were sharing a computer while they hurried to create Doom?
From the caption for the NeXT Cube photo in the linked gallery showing Tom Hall, it's clearly his machine:
This picture was taken during DOOM development in 1993.
Tom's busy working on a map in my map editor, DoomEd,
but don't worry - you won't be able to make out anything on
the screen! Tom started many of the original DOOM maps and
after he left, Sandy Petersen finished them -- some of the
maps had very little work done to them, in fact.
On Tom's monitor is a little printout that says, "Quality".
He always has notes stuck to his monitor to keep him
reminded of .... things. That orange thing could very well
be the very first keyboard wrist pad -- made by a friend of
his back before you saw them in stores. The speaker at the
far right is one of the first Altec Lansing clamshell
models. The microphone in front of Tom was there for
NeXTSTEP's awesome email feature - built-in voice emails! We
always sent them back and forth... I think I might have some
of them lying around here...
[1] https://romero.smugmug.com/Video-Games/The-Archives/i-T8gzwx...Edit:
It's possible they all moved onto NeXT machines after the success of Doom, but I am not aware of it. The NeXT tooling was not mature at the time and targeting x86 Windows/Dos from NeXT would have been a real nuisance. That's why it made sense for tools, since they could develop and run them on the NeXT machine.
Actually, I stand corrected, if Romero's account here is honest they appear to have done it all on NeXT. This is entirely inconsistent with what I read at the time:
https://web.archive.org/web/20070305165006/http://rome.ro/20...
> This picture was taken during DOOM development in 1993. Tom's busy working on a map in my map editor, DoomEd, but don't worry - you won't be able to make out anything on the screen! Tom started many of the original DOOM maps and after he left, Sandy Petersen finished them -- some of the maps had very little work done to them, in fact.
> On Tom's monitor is a little printout that says, "Quality". He always has notes stuck to his monitor to keep him reminded of .... things. That orange thing could very well be the very first keyboard wrist pad -- made by a friend of his back before you saw them in stores. The speaker at the far right is one of the first Altec Lansing clamshell models. The microphone in front of Tom was there for NeXTSTEP's awesome email feature - built-in voice emails! We always sent them back and forth... I think I might have some of them lying around here...
It's not even a page of copy/paste code to make a hook do whatever you want. Writing a .so for LD_PRELOAD is similarly a handful of lines.
Or you can use the plethora of open source debugging LD_PRELOAD libraries like efence.
And in a larger sense, you should try learning more and getting better at programming instead of being so dismissive.
Out of curiosity, what's something you consider difficult or at the edge of your abilities while programming C++ in Linux? (Maybe something related to code generation?) You're clearly a very advanced user.
Even something basic like replacing malloc with your own instrumentation to track errant allocations is basically impossible since every DLL in Windows loads its own allocator.
I know Windows has similar tools but wrangling Visual Studio, especially remotely or across multiple developers, is significantly more painful in my experience.
Even if you don't ship on Linux, programming on Linux is worth it imo. I mean, Carmack wrote Quake on a NeXT and shipped on Windows, so if he was doing it back in the day, how wrong can it be?