Hacker News Reader

Comment by bayindirh | Hacker News Reader

Yes, a lot of younger developers and "tech citizens" think MS is a saint now. They need to love Open Source / Linux because their platform has lost to the openness, composability and lightness of Linux.

They lost the server space, but they're retrying to seize desktop from Linux with WSL and invading the developer tools space and trying to isolate Desktop Linux as an "irrelevant layer which runs MS software and technologies, which is not visible and accessible as the laymen", and Linux server as a layer "which you don't need to think about" (i.e. make Linux irrelevant as a whole OS for developers at scale).

Canonical is helping them with this endeavor, because Ubuntu desktop doesn't make money. If they can piggyback the userspace to Windows and earn money from servers, they're kosher.

It's really ugly. However, De Icasa is not a saint either by pushing Mono on GNOME and Linux and allowing some borders to dissolve to allow Microsoft in, regardless of it's intended or not.

View on HN

> trying to isolate Desktop Linux as an irrelevant layer

They're obviously doing that, but they don't really need to.

Desktop Linux is its own enemy. 2 major GUI toolkits, frequent upgrades to them, generally with breaking changes, multiple DEs, frequent upgrades to them, also with breaking changes, a full decades of bickering over a solution that solves actual end user and operator problems in a standardizing way (systemd), immature GUI software after decades (Gimp & co), chronic lack of funding because nobody can figure it out, etc.

I've been around Desktop Linux for more than 15 years. I still hear bad arguments I heard back in 2005.

Plus, the ship has sailed. Mobile devices outnumber desktops and laptops and will only continue to grow. The money surrounding mobile dwarfs desktop. And Linux lost on mobile, it was it best chance to become the primary platform somewhere. Android doesn't count. Android used a cheap and standard kernel with many drivers and plopped a million proprietary things on top that look nothing like Desktop Linux.

You'll say I'm wrong, yet I'm willing to bet $100 that this discussion will look the same way in 15 years.

I'll just say that while Linux desktop is not perfect, I see much more people who're "assembling" their own environments to work at speeds I can't ever imagine.

I'm part of a community who does amazing things with all these fragmented parts and assemble unimaginable things while staying productive and inventing new ways of work.

The overall maturity of these small parts have not reversed the tide, but actually changed it a lot.

As a macOS and Linux user for 15 years, I see that both can work seamlessly without much effort, but Windows built on a completely different platform, can't talk this language of protocols and composability natively, or work as seamlessly.

Microsoft is right to feel threatened, because they're threatened by two sides.

> people who're "assembling" their own environments to work at speeds I can't ever imagine.

People who assemble cars from scratch can accomplish things from that are amazing too, but only 0 1% of the population will ever want to do that.

The thing is, some of these people are professional developers, and they inspire their peers and younger generations. There's an active momentum around that and it's definitely larger than 15 years ago.

It is 2022 and my suse linux smb share still doesnt work (after spending a few hours on it)

> Android doesn't count. Android used a cheap and standard kernel with many drivers and plopped a million proprietary things on top that look nothing like Desktop Linux.

Android is doing what every Linux embedded platform had been doing anyway. If anything, they're more open than the rest of the mobile/embedded space. It's not like e.g. OpenMoko looks more like Desktop Linux than Android does.

I think the goal of a single, quality "desktop linux" (or really a "free desktop", whatever kernel drives it), in the face of so much software diversity, is a lofty and improbable goal. The only issue probably is to match an ideal ("all IT departments use free software") with the practicality ("there is too much software diversity") with a fear that our current free software will fall out of favor with people seeking to work on and maintain software.

It is unfortunate though that some free competitors to commercial software still struggle a bit. GIMP has worked fine for my rather simple image editing needs though.

I think mobile is another beast, I'm not going there yet :D

> 2 major GUI toolkits, ... multiple DEs, frequent upgrades to them, also with breaking changes, a full decades of bickering

I was using Linux desktops 20 years ago and the same arguments were made. They were FUD and they still are.

1) having "too much" choice is nothing to complain about.

2) if you are a beginner, pick a stable distribution and your desktop will be completely immune to churn. Use it as a productivity tool, not as a playground.

3) if you are not a beginner you should be able to maintain a desktop. If you are busy, pick a simple DE and use it on a well know distribution.

> They were FUD and they still are.

Sorry, but take the horse blinders off.

If this kind of criticism coming from a long time user of Linux desktop (not as active nowadays because work is Windows, but most of my work is done in IDEs, browsers and through Cygwin anyway), FOSS supporter, etc is FUD, then boy, are you going to have a hard time fighting Microsoft's or Apple's marketing teams.

> 3) if you are not a beginner you should be able to maintain a desktop.

And therein lies the main rub :-)

Most people aren't able to "maintain a desktop". They don't really maintain either Windows, or MacOS. Most people are at best, "perpetual intermediates". Only a small fraction of users want or actually become power users. Most just trundle along trying their best to learn the least possible. And that's in every domain you can imagine.

This is the thing that will condemn Desktop Linux to being a niche product. That's fine, up to a point.

> Most people aren't able to "maintain a desktop".

He offered the solution though. If you are not able to "maintain a desktop" (either because you don't know how to, just don't want to, or don't have the time to, etc) then use a stable distro and accept the choices that were made for you. Thats not much different than just installing Windows or buying a Mac and accepting whatever decisions Microsoft/Apple made regarding how the desktop experience should be.

What part of "pick a stable distribution" you did not understand?

There's a good number of large tech companies providing Linux desktops for both tech-savy and other employees and it just works.

> Sorry, but take the horse blinders off.

Horse blinders? Ironic given how you looked only at half a sentence and ignored the rest.

Nothing like telling your boss "Sorry I missed the deadline, but my desktop needed maintenance"

Thats why he mentioned using a stable distribution as the default choice. Desktop maintenance is not required with a stable distro. It is for power users who have both the desire to customize their workflow and the time to configure and maintain it. Neither Windows nor macOS offers any suitable options for that type of user.

The idea of a desktop needing "maintenance" is something the industry has been moving away from.

Hence nr2?

> I'm willing to bet $100 that this discussion will look the same way in 15 years

I'm willing to take it, but only because I think there won't be a Linux in 15 years.

And yeah. Android doesn't count. It's essentially an enclave that runs Dalvik programs.

Linux basic assumptions about hardware don't hold on mobile.

I think Linux is suffering from the Lisp curse. Because they never choosed a DE, there is endless splintering and re-invention of the wheel.

> I'm willing to take it, but only because I think there won't be a Linux in 15 years.

Let's see. I've added a reminder in my calendar with a link to this comment.

> Because they never choosed a DE...

But they created a Desktop Standard which allows every DE to behave the same where it matters. This unified a lot of things. So every DE is a shell and workflow with same underlying principles underside. Much like POSIX.

> every DE to behave the same where it matters.

Sure. Except the end user won't care. As end user I don't see it implements DS.

I see that KDE has bug A and Gnome has bug B. Or KDE has feature C that Gnome doesn't have, and vice versa. All in all as user. No DE fits me because bugs and features are scattered around.

E.g. Haiku did it right. You don't get to choose. It's baked into the OS

> I see that KDE has bug A and Gnome has bug B. Or KDE has feature C that Gnome doesn't have, and vice versa. All in all as user. No DE fits me because bugs and features are scattered around.

This is a "grass is greener over the fence" perspective. You can install apps for missing functionality or report bugs for the things you don't like.

Except, Windows and macOS doesn't care about your bug reports, and some of the applications are not implemented on these systems because there's no financial incentive, and open source tools for these platforms are low quality.

On the other hand, this modularity and composability does wonders in the Linux side.

Linux DEs have feature levels. from LXDE to E17/XFCE to GNOME/KDE. People try, change, select and settle. Their most important settings are carried around. BTW, Qt/KDE guys allows amazing interoperability between GTK and Qt, regardless of your primary choice.

So, after some time it's borderline nitpicking IMHO.

> This is a "grass is greener over the fence" perspective.

Not really. Just because neighbors grass is greener, doesn't mean Linux grass isn't a black wilting mess.

> So, after some time it's borderline nitpicking IMHO.

Example of "nitpicking" I/colleagues ran into: Multi monitor support, HDR, DE bricking after update.

And due to Linux fractious nature even if one distro fixed it, another would still have that issue.

Stuff that never happened to me with Windows/MacOS.

> Not really. Just because neighbors grass is greener, doesn't mean Linux grass isn't a black wilting mess.

Linux's grass was a much darker green in the past, but wasn't black, neither back then, nor now.

> Multi monitor support

Just plugged a secondary monitor into one of my colleagues Linux box a week ago, a second desktop appeared three seconds later. Used two monitors for 6 years, until 6 years ago. It was working fine, then.

> HDR

Don't have an HDR panel, can't say a word about it, but VLC handles HDR-SDR conversion just fine.

> DE bricking after update.

My current MacBook Pro bricked after an OS update, because of a bug in an OpenVPN client's OS integration. No, the tool is paid, and no, macOS has no OpenVPN client embedded in it. Maybe I shouldn't have installed a paid, premium OpenVPN client, who knows. Anyway, I wiped and reinstalled the thing as a result.

The symptom? A frozen progress bar under the Apple logo. No time remaining, no error, nothing. Can I have the logs please? "Haha, of course not, you peasant."

And I only wiped and re-installed a Linux distro once, 15 years ago, because I did something the tool warned me to not to do.

And Windows somehow succeeds to corrupt my older computer's BIOS after installing updates and reboots. Sometimes it hangs all fans blazing during the restart. No, it's not the battery. That also never happened with Linux.

So, I arguably have a much better experience with Linux w.r.t. macOS and Windows in the last 15 years.

> Just plugged a secondary monitor into one of my colleagues

Sure. And I had encountered it - from just not registering, it being "there" despite being unplugged. And it randomizing windows on plugging in.

Using Linux always feels like fighting uphill battle with gremlins.

> Don't have an HDR panel, can't say a word about it, but VLC handles HDR-SDR conversion just fine.

Pity, really.

Neither Weyland nor X.org support it. Amazing, another fork in the road.

> My current MacBook Pro bricked after an OS update, because of a bug in an OpenVPN

That's different. This was post install update. Clean machine.

I ran into similar issue that bit Linus from. Linus Tech Tips.

I updated pop_OS! on fresh install and bang! Welcome console my old friend.

I remember Linux 15 years ago and it was even worse. But just because it got better doesn't mean it's good.

> I see that KDE has bug A and Gnome has bug B. Or KDE has feature C that Gnome doesn't have, and vice versa. All in all as user. No DE fits me because bugs and features are scattered around.

Can exactly the same not be said for Windows or MacOS? The difference being that you're basically locked into your choice with Windows/MacOS.

>I think Linux is suffering from the Lisp curse. Because they never choosed a DE, there is endless splintering and re-invention of the wheel.

But who is "they", Linux the OS doesn't exist as Linux is just a kernel. There are countless distributions that assemble operating systems from parts, throwing GNU userland on top of Linux kernel and other bits of software on top of the GNU userland and the kernel.

FreeBSD, NetBSD, OpenBSD are operating systems, the developers do both the kernel and the userland.

In the early days of PC BSD I've tried to kickstart a new DE, just for PC BSD to break free from the rest of open source operating systems, and to have an UI that only work on PC BSD, is consistent - the same way Windows and macOS have their UI layers. And I failed at it, because I am human and I need to earn a living.

> But who is "they"

People that wrote if not the original kernel. So l guess Linus and co.

If an OS is to "win" desktop, or mobile, or whatever, it needs to minimize forking between its libs. Go for one way to do it, rather than a muriad of related ways

In a market share sense, Linux doesn't and never existed in the desktop in first place. However it's ubiquitous in servers and that success has nothing to do with a DE. Its universal success in server space also means many more people are acquainted with Linux today than Lisp in its heyday. If you consider the current 2% market share in desktop to be "existing", I will posit that these people aren't going away just because of fragmented UI, because I believe bulk majority of them are spill over technical users who use Linux despite the state of DE.

I'm not referencing the market share.

I'm referring to Lisp Curse. http://www.winestockwebdesign.com/Essays/Lisp_Curse.html

In summary: Because Lisp was easy to fork, any time two nerds disagreed it was forked. And because they weren't working on same database you got Lisps with patchworks of functionality/bugs that each on their own couldn't compare to more or less monolith language like C/C++ or Java.

Same dynamic in Linux land.

So instead of having 1000 developers on one codebase we have 100 on ten different DE, windowing system, etc.

Rather than having one sound system we have a dozen... It doesn't work on PulseAudio, try OpenSound, no wait OSSBlasterMaster...

Scheme and Common Lisp are different environments altogether, Also, Emacs Lisp has a good chunk on fuctions backed it.

>s that each on their own couldn't compare to more or less monolith language like C/C++ or Java.

Good joke. On C++, Boost. Try to compile older C++ code from 2006 today, it may fail spectacularly.

A lot of older C++ games from 2006 will be damn difficult to compile today.

On Java, heh. Different and incompatible JRE's. Software tied to a single JRE where it crashes on the newer ones.

Should I go on?

Meanwhile, 20 yo Elisp code still runs today as-is.

Yup, agreed. Didn't realize how much Canonical was helping with WSL, though it of course makes sense that they're doing so, unfortunately.

> De Icasa is not a saint either by pushing Mono on GNOME and Linux

Eh, I think interoperability is good, and reverse-engineering Windows frameworks/technology has a long history in the open-source world (WINE & Samba being two prominent examples). I was disappointed when they sold Xamarin to MS, though. That felt like selling out, even though I'm sure it made sense at the time.

> Didn't realize how much Canonical was helping with WSL, though it of course makes sense that they're doing so, unfortunately.

Eh, they have a dedicated team for enabling WSL/Ubuntu desktop in Canonical. They're fueling the effort.

> Eh, I think interoperability is good, and reverse-engineering Windows frameworks/technology has a long history in the open-source world (WINE & Samba being two prominent examples).

IIRC, De Icaza left GNOME, joined Microsoft, developed Mono on open specs, founded Xamarin and sold it back to Microsoft.

His joint venture with MS what turned everything sour for many people, incl. me. However, we all love Wine, esp. when they implement things bug-by-bug without knowing it (and without help from Microsoft). The cursor (or Icon or WMF) exploit was a great laugh. We all love SMB too, because it's done despite Microsoft.

The language on this page is loaded (in a way I don't like), but it keeps a good roster of the events: http://techrights.org/wiki/Miguel_de_Icaza

> His joint venture with MS what turned everything sour for many people, incl. me.

Why did it turn things sour for you? I mean, ever since the Xamarin acquisition, MS .NET has become open + cross plattform and Xamarin which was a commercial/paid-for product is open too. VSCode is a first class citizen for free C#/.NET development on macOS and Linux. Not saying all of this is related to de Icaza getting bought by MS, but just saying ever since the ecosystem has become completely open.

If you're a .NET developer who uses Linux, yes that sounds nice.

On the other hand. Consider a prominent member one of the biggest open source projects leaves the project, joins Microsoft, and praises Microsoft and pushes semi-closed Microsoft technologies on every project, claiming Linux is dead without Microsoft, and we need to compromise by allowing these non-free technologies and porting almost everything to it.

All is done when Microsoft tried to copy LibreOffice (then OpenOffice) formats, created OOXML family of semi-open formats, and push them as ISO standards to bolster their file formats to kill other office applications for good.

DeIcaza (knowingly or unknowingly) was a part of an concentrated and coordinated effort to EEE Linux, and the open frameworks around it.

I never used and won't use any Mono based application in my personal systems and servers I solely install and manage.

On the other hand, VSCode is not free or open source. This makes it null and void in my radar, because I want my whole toolchain to be open and free, because when I left a project alone, someone should be able to get the whole toolchain and continue development. I don't like the idea of source code being locked in a proprietary development environment and hence suffocated on a whim with a choice of an external party.

> DeIcaza (knowingly or unknowingly) was a part of an concentrated and coordinated effort to EEE Linux, and the open frameworks around it.

The only person who does not know this is Icaza itself.

I can't talk about a person definitively who I don't know personally.

VSCode is free software, and also used a lot outside .NET development (TypeScript). It is extension based, and not all extensions every written for it are free software. And VSCode is not essential in any way for .NET development, and I am saying this as someone who learned C# with MonoDevelop on Linux around 2010. There are commercial IDEs like Rider, and you don't need an IDE to compile code - the build toolchain of .NET is free software.

I don't buy into this "All tools associated with an ecosystem must be free software". That is not even true in Java/C++/Python land, where there is a plentheora of commercial and succesful developer tools (i.e. IntelliJ, CodeWarrior etc.)

> VSCode is free software...

As in beer only [0]. This license [1] states otherwise for free as in speech perspective.

> and also used a lot outside .NET development (TypeScript). It is extension based, and not all extensions every written for it are free software.

I know, and some of the non-free extensions are just the essentials, so they're what make VSCode, VSCode. So it's open-core, and it's not completely open source, and definitely not free.

> And VSCode is not essential in any way for .NET development,

I don't develop C#, but I know there are other tools and IDEs.

> I don't buy into this "All tools associated with an ecosystem must be free software". That is not even true in Java/C++/Python land, where there is a plentheora of commercial and succesful developer tools (i.e. IntelliJ, CodeWarrior etc.)

It may not be a necessity for you, and it's definitely not a necessity for closed source software or tools. However, I license all my code with GPLv3, and want my project to be buildable by anyone, regardless of anything. So I prefer a free and open toolchain (GCC/Eclipse/GNU libc) and libraries in my codebases.

If I was writing a commercial application, I may make different choices IDE/Library wise, but I'll still continue to use GCC family.

You don't have to buy into this. I'm not selling anything. This is my stance and a recommendation, but not an insistence.

[0]: https://code.visualstudio.com/docs/supporting/faq#_is-vs-cod...

[1]: https://code.visualstudio.com/license

The .NET debugging tools in VS code are proprietary. If you get the OSS build from your distro and try to debug a .NET program it'll fail with an error. (Telling you it's only available with proprietary VS code from MS) For other languages this might be a minor pain. For C# it kills one of its best features, it's debugging support.

This is also an ecosystem thing. Rust has pervasive support for printf debugging (#[derive(Debug)] and you're done), while .NET's default .ToString() method is completely useless (which I'm sure has nothing to do with Microsoft wanting people to rely more on their debugger...).

Yeah you don’t remember correctly at all.

This is why I added Techrights reference here, which fixes what I've been wrong about and supports me what I was right about.

The most important thing is, De Icaza is with Microsoft for much longer than people remember (or want to accept), and the events I remember (albeit a bit blurry), when I was a grad student were actually happened.

> even though I'm sure it made sense at the time

I think it definitely made sense. Xamarin was conceived in a world of many mobile platforms, and so it was the only way to develop cross-platform apps. Now there are basically two, and it's a much closer call.

Also stiff competition from React native. A lot more JS Devs around today than c#.

Yes true, but not all React ones :) And Unity has helped swell the C# ranks, I hear.

Reply on news.ycombinator.com