linux desktop in trouble

linux desktop in trouble

[Mehdi Saada]

https://www.zdnet.com/article/the-linux-desktop-is-in-trouble/
How easier would it be if they weren't using messed-up tools to begin with, if they had the intelligence to let compilers do more of the work for them ?
How much less work hours in maintenance and errors correcting ?

Re: linux desktop in trouble

[Leif Roar Moldskred]

Mehdi Saada <00120260a@gmail.com> wrote:
> https://www.zdnet.com/article/the-linux-desktop-is-in-trouble/
> How easier would it be if they weren't using messed-up tools to begin with,

Not a lot, honestly. The problems that plague the linux desktop are
political and architectural, not issues of functional correctness or bug-
fixing.

-- 
Leif Roar Moldskred

Re: linux desktop in trouble

[Shark8]

On Tuesday, March 3, 2020 at 2:45:40 AM UTC-7, Mehdi Saada wrote:
> https://www.zdnet.com/article/the-linux-desktop-is-in-trouble/
> How easier would it be if they weren't using messed-up tools to begin with, if they had the intelligence to let compilers do more of the work for them ?
> How much less work hours in maintenance and errors correcting ?

It seems to me that a lot of the problems in that article stem from design. Take installation: the article says "but under the surface, there are half-a-dozen different ways to install programs." — The proper solution here is to provide at least a proper interface for install/uninstall, like Ada does for allocaters, or possibly a whole subsystem.


In terms of programming languages, the problems cited in that article are simply the result of the C mentality: prioritizing simplicity over correctness or completeness. — This is opposed to the Ada mentality of doing things foremost correctly which, arguably, could be the theme of several older OSes like the Burroughs MCP, Multics, and perhaps OpenVMS.

In order to really solve these problems, you would have to design the OS (from the ground up) to address these issues; IIRC some of the above-mentioned OSes did so, at least for various subsystems. (OpenVMS, for example, had a very nice networking subsystem [based on the OSI model, IIRC] where client-programs were unaware of the network-connection protocols.)

Re: linux desktop in trouble

[Optikos]

On Tuesday, March 3, 2020 at 4:14:27 AM UTC-6, Leif Roar Moldskred wrote:
> Mehdi Saada <00120260a@gmail.com> wrote:
> > https://www.zdnet.com/article/the-linux-desktop-is-in-trouble/
> > How easier would it be if they weren't using messed-up tools to begin with,
> 
> Not a lot, honestly. The problems that plague the linux desktop are
> political

Politics polices policy (both in tech and in government).  The so-called “political” trouble with Linux in particular (and somewhat so in the BSDs) is that it is difficult to herd the cats to agree on one cohesive set of •policies•.  When agreement on the fundamental good versus bad of various policies cannot be achieved, then competing political factions arise to defend their favored set of policies (e.g., systemd versus some other init; C-centric GTK versus C++-centric Qt; pacman versus apt versus snap; Arch distro's rolling release versus Debian distro's periodic release versus Yocto-primordial-not-a-distro roll-your-own-distro).  Then each faction laments & lambastes too much politics when some political movement naturally organically arises that backs a different set of policies.

> and architectural,

1) Cathedral of perhaps never-ending refinement of perfection (which risks bicycle shedding)
versus
2) bazar of perhaps the just-barely-good-enough bizarre (which risks ossifying too soon in local optima that are far far from global optima)
are 2 competing sets of architectural policies that have 2 strongly-held politics policing each of them for decades.

> not issues of functional correctness

The Linux community as a whole cannot agree to one all-the-wood-behind-one-arrow* intended functionality.  Hence the Linux community as a whole cannot enforce perfect fidelity to one set of functionality because of 2 or more competing sets of functionality.

> or bug-fixing.

In the bazar, one person's bug often can be another person's desirableness in certain use-cases.

The answer to this is of course for the cathedral Ada community eventually to have a separate cathedral OS that is separate from the Linux bazar.

Re: linux desktop in trouble

[ldries46]

Op 3-3-2020 om 10:45 schreef Mehdi Saada:
> https://www.zdnet.com/article/the-linux-desktop-is-in-trouble/
> How easier would it be if they weren't using messed-up tools to begin with, if they had the intelligence to let compilers do more of the work for them ?
> How much less work hours in maintenance and errors correcting ?
As a newcomer in the Linux world I can feel what he means. But I just 
think that it is not so much the problem of the desktops themselve but 
of the lack of standardized installation procedures. There should be 
some installation programs as there are in windows, these programs 
should be independent of the type of desktop you are using. And of 
course they should be complete.
Take for instance GNAT GPS and GtkAda.
GNAT GPS before running the installation you must make the package 
executable, it should be delivered execuable as under Windows. In Ubuntu 
you need to create a .Desktop File to run it from the desktop, that 
should be automaticely be created.
GtkAda in Windows is one package completely installable. Under Linux it 
is broken into several parts which you first have to locate.
This is not a problem of the desktop itself but of the producers of 
software.

Re: linux desktop in trouble

[Dmitry A. Kazakov]

On 2020-03-19 07:07, ldries46 wrote:

> As a newcomer in the Linux world I can feel what he means. But I just 
> think that it is not so much the problem of the desktops themselve but 
> of the lack of standardized installation procedures. There should be 
> some installation programs as there are in windows, these programs 
> should be independent of the type of desktop you are using. And of 
> course they should be complete.

Well, it is rather exactly the opposite. Windows has no standardized 
installation procedure and rules, at all. There exist msi files, very 
few use them and it does not make much difference. Otherwise, it is just 
an executable that does any wild thing it pleases after you allow it to 
elevate itself to the administrative rights.

> Take for instance GNAT GPS and GtkAda.

It is especially so with GTK. Basically any Windows program that uses 
GTK hoards its own copy of. It is a total mess under Windows.

> GNAT GPS before running the installation you must make the package 
> executable, it should be delivered execuable as under Windows.

No. It simply should be better maintained as a deb-package, if we are 
talking about Ubuntu.

Situation under Linux is far from perfect but is way better than under 
Windows.

> This is not a problem of the desktop itself but of the producers of 
> software.

True. It would be nice if AdaCore packaged its community GNAT for major 
Linux platforms: Debian, Ubuntu, Fedora, CentOS. The problem is the 
resources. It is a lot of work, which, I assume, they prefer to apply on 
other places.

FSF packages are simply undermaintained. The compiler itself works out 
of the box. But GtkAda, GPS (and frequently gprbuild) always had problems.

-- 
Regards,
Dmitry A. Kazakov
http://www.dmitry-kazakov.de

Re: linux desktop in trouble

[Optikos]

On Thursday, March 19, 2020 at 3:46:48 AM UTC-5, Dmitry A. Kazakov wrote:
> On 2020-03-19 07:07, ldries46 wrote:
> 
> > As a newcomer in the Linux world I can feel what he means. But I just 
> > think that it is not so much the problem of the desktops themselve but 
> > of the lack of standardized installation procedures. There should be 
> > some installation programs as there are in windows, these programs 
> > should be independent of the type of desktop you are using. And of 
> > course they should be complete.
> 
> Well, it is rather exactly the opposite. Windows has no standardized 
> installation procedure and rules, at all.

Wrong!  There are a plethora of extraordinarily strict rules that an app (either desktop or UWP) must obey to be admitted into the 1 official app distribution infrastructure:  Microsoft Store.

https://docs.microsoft.com/en-us/windows/uwp/publish/store-policies-and-code-of-conduct

Even if an x86/x64-only desktop app seeks to distribute through the Microsoft Store and forego ARM, Hololens, and Xbox deployments:
https://developer.microsoft.com/en-us/windows/bridges/desktop

Re: linux desktop in trouble

[Randy Brukardt]

"Optikos" <ZUERCHER_Andreas@outlook.com> wrote in message 
news:b11b23e7-fec1-4deb-ba63-2cb0b4e15f5c@googlegroups.com...
> On Thursday, March 19, 2020 at 3:46:48 AM UTC-5, Dmitry A. Kazakov wrote:
>> On 2020-03-19 07:07, ldries46 wrote:
>>
>> > As a newcomer in the Linux world I can feel what he means. But I just
>> > think that it is not so much the problem of the desktops themselve but
>> > of the lack of standardized installation procedures. There should be
>> > some installation programs as there are in windows, these programs
>> > should be independent of the type of desktop you are using. And of
>> > course they should be complete.
>>
>> Well, it is rather exactly the opposite. Windows has no standardized
>> installation procedure and rules, at all.
>
> Wrong!  There are a plethora of extraordinarily strict rules that an app
>(either desktop or UWP) must obey to be admitted into the 1 official app
>distribution infrastructure:  Microsoft Store.

Sure, but does anyone actually use that for desktop software? I've never 
once gotten anything from it.

Note that Microsoft has had such requirements for decades (for instance, to 
get a "logo"), but those have been widely ignored. Especially for cheap/free 
software and highly speciallized software.

I believe I spent the time to read them once. ;-) That's it. Microsoft sets 
requirements to make Microsoft money; little actually helps the users (and 
when they accidentally land on something that does help users, they change 
it in a year or to because someone had a better idea).

                                    Randy.

Re: linux desktop in trouble

[Dennis Lee Bieber]

On Fri, 20 Mar 2020 15:43:25 -0500, "Randy Brukardt" <randy@rrsoftware.com>
declaimed the following:


>
>Sure, but does anyone actually use that for desktop software? I've never 
>once gotten anything from it.
>

	Probably doesn't count as "desktop software" since it only supports a
command line interface... I've installed the Debian (and Ubuntu for some
reason) "Linux on Windows" packages via the M$ Store <G>



-- 
	Wulfraed                 Dennis Lee Bieber         AF6VN
	wlfraed@ix.netcom.com    http://wlfraed.microdiversity.freeddns.org/

Re: linux desktop in trouble

[mgr]

El 19/3/20 a las 7:07, ldries46 escribió:
 > As a newcomer in the Linux world I can feel what he means. But I just 
think that it is not so much the problem of the desktops themselve but 
of the lack of standardized installation procedures. There should be 
some installation programs as there are in windows, these programs 
should be independent of the type of desktop you are using. And of 
course they should be complete.

You are right, but that gap is starting to be filled by snaps 
(https://snapcraft.io/) or AppImage files.

 > Take for instance GNAT GPS and GtkAda.
 > GNAT GPS before running the installation you must make the package 
executable, it should be delivered execuable as under Windows. In Ubuntu 
you need to create a .Desktop File to run it from the desktop, that 
should be automaticely be created.
 > GtkAda in Windows is one package completely installable. Under Linux 
it is broken into several parts which you first have to locate.
 > This is not a problem of the desktop itself but of the producers of 
software.

They should consider using a snap package and the process would be easy 
as installing a distribution package.

Re: linux desktop in trouble

[Bob Goddard]

On Thursday, 19 March 2020 06:07:21 UTC, ldries46  wrote:
[...]
> Take for instance GNAT GPS and GtkAda.
> GNAT GPS before running the installation you must make the package 
> executable, it should be delivered execuable as under Windows. In Ubuntu 
> you need to create a .Desktop File to run it from the desktop, that 
> should be automaticely be created.
[...]

Eh, part of the windows problem is that files can be executed just by their name. Under Unix/Linux, you do have better control.

And no, you do not need to create .Desktop file to run it. From a shell, chmod +x <file>, then ./<file>...

Re: linux desktop in trouble

[ldries46]

I do not agree with Dmitry for the simple reason that every installing 
routine asks from the sender take amateures into account. and as Gtk is 
that is as I see it more a problem for the designers than of the 
operating system itself. Even under windows you can make a mess of it.

Op 19-3-2020 om 9:46 schreef Dmitry A. Kazakov:
> On 2020-03-19 07:07, ldries46 wrote:
>
>> As a newcomer in the Linux world I can feel what he means. But I just 
>> think that it is not so much the problem of the desktops themselve 
>> but of the lack of standardized installation procedures. There should 
>> be some installation programs as there are in windows, these programs 
>> should be independent of the type of desktop you are using. And of 
>> course they should be complete.
>
> Well, it is rather exactly the opposite. Windows has no standardized 
> installation procedure and rules, at all. There exist msi files, very 
> few use them and it does not make much difference. Otherwise, it is 
> just an executable that does any wild thing it pleases after you allow 
> it to elevate itself to the administrative rights.
>
>> Take for instance GNAT GPS and GtkAda.
>
> It is especially so with GTK. Basically any Windows program that uses 
> GTK hoards its own copy of. It is a total mess under Windows.
>
>> GNAT GPS before running the installation you must make the package 
>> executable, it should be delivered execuable as under Windows.
>
> No. It simply should be better maintained as a deb-package, if we are 
> talking about Ubuntu.
>
> Situation under Linux is far from perfect but is way better than under 
> Windows.
>
>> This is not a problem of the desktop itself but of the producers of 
>> software.
>
> True. It would be nice if AdaCore packaged its community GNAT for 
> major Linux platforms: Debian, Ubuntu, Fedora, CentOS. The problem is 
> the resources. It is a lot of work, which, I assume, they prefer to 
> apply on other places.
>
> FSF packages are simply undermaintained. The compiler itself works out 
> of the box. But GtkAda, GPS (and frequently gprbuild) always had 
> problems.
>

Re: linux desktop in trouble

[Dmitry A. Kazakov]

On 2020-03-24 11:01, ldries46 wrote:

> I do not agree with Dmitry for the simple reason that every installing 
> routine asks from the sender take amateures into account. and as Gtk is 
> that is as I see it more a problem for the designers than of the 
> operating system itself. Even under windows you can make a mess of it.

Windows simply has no means to handle software components coming not 
from Microsoft. So GTK's run-time and/or developing cannot be versioned 
and shared by applications. It is not GTK problem it is Windows problem. 
E.g. Python has exactly same issues.

-- 
Regards,
Dmitry A. Kazakov
http://www.dmitry-kazakov.de

Re: linux desktop in trouble

[Optikos]

On Tuesday, March 24, 2020 at 5:29:08 AM UTC-5, Dmitry A. Kazakov wrote:
> On 2020-03-24 11:01, ldries46 wrote:
> 
> > I do not agree with Dmitry for the simple reason that every installing 
> > routine asks from the sender take amateures into account. and as Gtk is 
> > that is as I see it more a problem for the designers than of the 
> > operating system itself. Even under windows you can make a mess of it.
> 
> Windows simply has no means to handle software components coming not 
> from Microsoft.

Wrong!  Package dependencies can be declared in Windows 10, then utilized at time of installation of (nonMicrosoft) packages that depend on other (nonMicrosoft) packages:
https://docs.microsoft.com/en-us/uwp/schemas/appxpackage/uapmanifestschema/element-dependencies

“Dependencies must be explicitly defined. If a dependency cannot be resolved, deployment of the package fails.”

> So GTK's run-time and/or developing cannot be versioned 
> and shared by applications. It is not GTK problem it is Windows problem. 
> E.g. Python has exactly same issues.
> 
> -- 
> Regards,
> Dmitry A. Kazakov
> http://www.dmitry-kazakov.de