A counter-example doesn't disprove a general point.
You're describing a strong tie-in in your work environment to specific commercial software articles that are unavailable on Linux. This does not characterize most people (or even a large enough minority).
Now, you could make the argument that people interact with others using MS-Office documents, and that support for them in LibreOffice is insufficient. One could argue this both ways (as support has improved over the years and is by now passable IMHO), but that's not the same as what you're using.
In general, it _is_ true that people use the Operation System that came installed on their system. Most people are unable, or feel unable, to install an OS themselves and would not feel comfortable taking responsibility for choosing a different OS for their computing. Most do not even see this as a choice they are making.
Finally, the "foreign ecosystem" argument is circular. If you get a computer preinstalled with some operating system, and you learn how to use that, than other systems seem foreign. Few people get a Mac as a present then try to install Windows on it because it's a "foreign ecosystem"
> Doing a fully functional ports of software people use under Windows would've
It's the commercial companies which sell this kind of software that can port it.
This suggests that, if given the option, people would buy a computer with Linux preinstalled instead of Windows. Yet, when such things were tried, people did not. Windows is pre-installed because it is the OS people want and need, not the other way around. There are an uncountable number of specialized applications for niche workflows available on Windows that are not available on Linux.
As briefly mentioned by the parent, a large part of the problem is how Linux approaches software: no real binary compatibility, for instance, means that you need this army of maintainers and packagers to keep software working. Who is going to do that for these niche pieces of software? I know that to many developers the idea of not-constantly-maintained software being used by people is an existential career threat and therefore a high crime, but the rest of the world has work to do and is totally fine using VB6 applications last compiled in 2004 because it allows them to actually get things done.
No, what I said does not suggest that.
> no real binary compatibility, for instance, means that you need this army of maintainers and packagers to keep software working.
It's a smaller army than the army you need in order to keep Microsoft Windows working, and even smaller compared to the army you need to keep the Windows software equivalent to what you get as part of a Linux distribution.
> There are an uncountable number of specialized applications for niche workflows available on Windows that are not available on Linux.
That's not relevant to my argument.
I consider myself a geek, a fairly advanced user of software and a software developer.
I have been trying different versions of Linux since the Mandrake Linux times, so about what, 20 years?
I always come back to Windows, which from Windows 2000 has been getting better and better. Of course there have been sh*t moments like Windows Vista and Windows 8, but Windows 2k, XP, 7, and now 10 (enterprise version), have always been good to me.
I can use Linux for sure, and I appreciate the effort the community does, however, for my use case, I don't see the advantage of using Linux over Windows, and I find many disadvantages: missing software (office, adobe, although I have run it under wine), missing or not perfect drivers (energy efficiency is a problem even on thinkpads).
Again, I really like the effort, and if I must use it, I can, but I don't see the advantage.
Now, I use it on my dad's computer (he is 80), as it is much difficult for him to screw up than Windows.
For me though, software isn't missing on Linux, it's missing on Windows. I rarely need an office suite, and my needs are limited to simple Word/Writer documents and even simpler Excel/Calc spreadsheets, so even MS Office 97 or StarOffice 5 have all the needed features (minus support for XML file formats). The extent of image editing I do is limited to cropping, rotating and resizing images, so even GIMP is way overkill.
On the other hand, the software I do use and like is a pain on Windows. First it's basic text manipulation tools. Things like grep and sed are missing on Windows, and I want to e.g. "replace this line in all files in this folder" far more often than I open an Excel file. I use SSH, which isn't on Windows, and applications like putty are great but don't provide anywhere near the seamless experience that Linux has with SSH. Then there's the desktop environment itself - I use KDE Plasma 5 and find it amazing, having been a fan of KDE since 3.5. I like KDE Plasma overall much more than Windows Explorer, and then there are the individual power features, e.g. I have a button on the titlebar that toggles always-on-top for a window, and I exclude certain applications from appearing in the taskbar because they already have a systray icon.
I use Linux at work, and haven't had to boot Win7 at home since Steam Proton. I've heard that Win10 makes some things better, like it has built-in SSH. But I haven't tried Win10 myself, and at this point it'd take some Windows killer feature to make me switch.
That said, for my use case, I have never found a software that I need on Windows that is only available on Linux.
The opposite happens to me a lot.
For the sake of fairness, it should be noted that most Linux software is open source and can be run on other platforms even if mainly meant for Linux. Git is a prime example, developed as a Linux program - in fact developed by Linus Torvalds specifically for the kernel - but it runs on Windows with a few quirks. LaTeX is a Linux-first project but you can also use it on Windows, etc.
This mostly doesn't happen in the opposite direction because of closed, proprietary software. Could Photoshop be ported to run on Linux, and with how much effort? We do not know.
> grep and sed are missing on Windows,
cinst grep sed dngrep ...
> "replace this line in all files in this folder"
Everything file search engine is beyond anything LInux has to offer.
> I use SSH, which isn't on Windows,
cist mls-software-openssh
Besides, Open SSH is now OTB in Windows
> use KDE Plasma 5 and find it amazing, having been a fan of KDE since 3.5. I like KDE Plasma overall much more than Windows Explorer,
All GUIs everywhere are equal SHIT. Its just your preference.
>OpenSSH has been added to Windows as of autumn 2018, and is included in Windows 10 and Windows Server 2019.
"Average User" is a strawman. The average human has fewer than two legs, yet we do not design things with the assumption that no one has two legs.
- Photoshop : your lifetime license is not really lifetime, CS2 servers got shut down recently, Adobe will shut down those servers sooner or later. While you may not find it true for you, for the non professional Gimp is generally good enough. Also the Mac version of Photoshop while having some kinks even for professional use is pretty comparable, you don't need Windows given you are already using a macOS.
- Enterprise Architect: If your argument is nobody is investing in Linux desktop modelling apps it can be argued nobody is investing in desktop modelling apps today period whether in Windows or Linux or Mac .
Modelling applications on the web/mobile like Miro(just raised $100M+) or draw.io or many others have gained a lot of traction. Sharing and collaborating on the model has lot more value than few advanced features, Yes today Enterprise Architect has more features, however that is not going to last and these apps have a lot of functionality already that average users will not find them lacking even today.
- MS Office: Sharepoint/o365 is almost as good as desktop office for most common use cases and getting better lot faster than upgrades to Excel. Yes Excel is used for those million row sheets with gazillion circular dependent formulas, and it works pretty darn well for all that people throw at it. Sharepoint is improving, the use cases it is poor at handling will become smaller and smaller that procurement departments will start questioning desktop license budgets for Office apps.
On the whole barring some exceptions like gaming, graphics etc there are only few new active investments for desktop apps that are not just electron apps of web.
OS is becoming irrelevant for productivity apps outside of these sectors and legacy reasons. It is not big tech does not know that, Apple is dropping dual boot in the next gen chip and it is not like windows or linux cannot support ARM, the pace of OS development has slowed considerably, Windows is only do to incremental updates to 10, the cost and effort of doing a major release is no longer worth the returns.
And again, I emphasized "software of that class" for a reason, because migrating "easy stuff" like productivity tools to Linux/Mac was not a hassle. I'm sure there's ton of new promising modeling solutions, but I can't stop using Enterprise Architect because of literal vendor lock. I use automation and I use code generation, which is implemented through .NET API (not .NET Core), which assumes Windows. However, I suppose, 99% of EA users don't worry it's Windows-only, for many enterprise-class solutions OS is just a wrapper for the software.
I am not trying to convince you of anything, and you have the right to hold to your contradictions, but I find this at odds with most proprietary software, even more so if their execution environment isn't stable or 100% under control.
Regarding your specific needs, it also does sound a bit like the XY problem: you don't really need an app (unless locked-in by data you can't migrate), you just need the functionality it provides, though that may require some unlearning first. Oh and yeah, of course there aren't alternatives for everything.
But if you are really locked in and see the writing on the wall for your solution, you always have the option of creating your own migration path, or a stub application that does the bare minimum you need. Other might step in and help you, so it might be worth asking around first.
From my point of view, Linux is not broken at all. It runs the software I need (I'm comfortable with Gimp, I don't have spreadsheets that pass the limits of LibreOffice and I don't use EA). Lately, the move from X to Wayland has been introducing some brokenness - it's not as easy as it used to be to run X apps across the network - but I imagine all these will be eventually fixed.
Like you said, most users don't worry EA is Windows-only because most users don't worry about things like OS.
My needs are far from average though, as I mentioned in the post. Not sure its spreadsheet editor would even recognize what's happening in my Excel files.
That is not quite true. At least Visual Paradigm exists for Linux. I consider it to be even better than Enterprise Architect.
https://www.visual-paradigm.com/download/index.jsp?platform=...
What did you miss in VP?
No compatibility with Windows, hardly anyone changes OS. Have not heard much complains about MS Office, Photoshop or Enterprise Architect. People use browser 99% of the time.
That's just a stereotype, and that's not true. I keep using Windows (hardened and almost isolated Parallels instance of Windows 7), because Linux has serious lack of software I need.
I need fully functional MS Office, including API bindings for data import to Excel, and no, Libre Office Calc does not have even 20% of features I need. Of course, I need 3rd party addons, like Grammarly for MS Office.
I need Enterprise Architect, modeling software of that class simply does not exist on Linux.
I need Photoshop (I have one since old good times when Adobe offered lifetime license), I use it occasionally and not 100% effective, but nope, there are no functional open-source alternatives for Linux with the same comfort of use.
People don't use Linux not because it is not pre-installed, they don't use it because it's a totally foreign ecosystem. Doing a fully functional ports of software people use under Windows would've probably moved 1% share of Linux on desktops, but it does not happen, partly because open-sourse software designers do not recognize the problem, partly because it's an immense amount of work, impossible without huge financial and human investments with unclear outcome.
I don't even say about technical issues, like total lack of backward compatibility, it's mostly a headache for techies like us.