It is not a terrible deep insight that wind is intermittent power source. It is also not new that there is less wind in the summer. On the other hand, there is more solar and the total production from renewables is surprisingly stable: https://www.energy-charts.de/r... Always about 35% which in summary is much more than nuclear.
Your charts highlight another point: The renewables clearly cut into consumption of coal and lignite. This is very good.
And there is another thing to learn: Nuclear is always running at the same level. This is related to a disadvantage of nuclear: It can *only* be used for baseload.Technically, it might be possible to do load following, but that makes the economics of nuclear even worse as you have very high investments you need to recover. On the one hand, this makes it very unattractive to have a lof of nuclear in a modern grid with intermittent power sources. On the other hand, you cannot use only nuclear because of demand side changes. So there is no real long-term use case for nuclear anymore.
I am not arguing against spending money to combat climate change, but the money we have available to combat climate change needs to be spent at cost effective solutions - and this means: not nuclear but wind, solar, storage, energy demand management, and more efficient technologies which reduce consumption. All these a much better use of the resources we have than subsidizing the nuclear industry which never delivered on their promises of cheap, safe, and reliable power.
Then I'd like to hear what is the answer. Think quickly because the clock is ticking.
The answers are renewables, storage, energy demand management, and saving.
It was promised to be too cheap to meter, instead it is the most costly to generate.
That's kind of irrelevant now, no?
No it is not. In a world of limited resources, you can not afford to waste money on an expensive technology as it reduces the money you can spend on more efficient solutions.
That might have something to do with those unwashed hippies that have been trying to "save the planet". The planet's fine, it's us humans that are fucked if we don't do something.
Unwashed hippies don't have much political influence. So no.
The only plant currently being built (Georgia) is 5 years behind schedule and double its original cost.
Then throw some more money at it. The alternative is potential extinction.
The alternative is to spend to money on more reasonable solutions.
This is electricity generation in Gemany from renewables from 1900 to 2017 in TWh / year: 19,7 25,1 37,9 38,9 46,1 46,1 57,2 63,1 72,4 89,1 94,1 95,7 105,2 123,6 143,3 152,5 162,5 188,6 189,8 218,3
This is substantial amount of power. At the same time price for renewables dropped, wind is basically competitive (hydro has been for long time) and for photovoltaics it is just a matter of time as price drop exponentially: https://en.wikipedia.org/wiki/...
In contrast, the huge amount of investments in nuclear which were ongoing for more than half a center did not achieve something remotely similar. It is still an inefficient technology which cannot compete without subsidies.
This is power generation in Germany from coal and lignite from 1990-2018 in TWh. coal 140,8 147,1 143,1 138,4 134,6 146,5 140,8 134,1 137,9 142,0 124,6 107,9 117,0 112,4 116,4 127,3 118,6 117,7 112,2 92,6 lignite 170,9 142,6 148,3 154,8 158,0 158,2 158,0 154,1 151,1 155,1 150,6 145,6 145,9 150,1 160,7 160,9 155,8 154,5 149,5 147,5
Also the CO2 emission from electricity production decreased from 315 Mio t CO2 emission in 2010 (before shutting down a couple of nukes in response to Fukishima) to 285 Mio t CO2 in 2017. At the same time power production increased from 564 TWh to 583 TWh.
Well, you stated that CO2 emissions increased in Germany since shutting down nuclear power plants. Technically this may be true, but this is highly misleading as CO2 emissions from electricity generation in Germany actually *decreased* and the overall increase (it it indeed exists for this time frame, I haven't checked) from CO2 comes from elsewhere (likely transportation). The CO2 emission from electricity production are 315 Mio t CO2 emission in 2010 (before shutting down a couple of nukes in response to Fukishima) to 285 Mio t CO2 in 2017. At the same time power production increased from 564 TWh to 583 TWh. So Germany decreased CO2 emissions from electricity production even after shutting down several nukes and increasing output. Whatever argument you tried to make, it based on a wrong premise.
The US is one of the countries with the highest per capita CO2 emissions per year substantially higher than Germany, Japan, China (and a lot higher than India.) This while having a trade deficit. No, we do not expect to pay for all other nations, we expect that you bring down *your* emissions.
Electricity generation has steadily reduced CO2 emissions in Germany. This has been offset by an increase in other sectors (transportation mostly). Making an argument against wind/solar is completely misleading.
This is power generation from coal and lignite from 1990-2018 in TWh. coal 140,8 147,1 143,1 138,4 134,6 146,5 140,8 134,1 137,9 142,0 124,6 107,9 117,0 112,4 116,4 127,3 118,6 117,7 112,2 92,6 lignite 170,9 142,6 148,3 154,8 158,0 158,2 158,0 154,1 151,1 155,1 150,6 145,6 145,9 150,1 160,7 160,9 155,8 154,5 149,5 147,5
If I have to first go to a mailing list of my distribution to learn about some random "reasons" to why some change is supposed to be good than this is almost by definition "no good reason". Is an actual improvement noticeable to the user too much to ask for?
I am scientist. I have to learn new stuff every day. I develop new stuff every day. But I have no sympathy for people wasting my time by breaking standard tools or conventions with no good reason. And the "you are just to lazy to learn new things" argument is just BS. I want to spend my time learning interesting things and not have to relearn how to do basic stuff with my computer because some random dude at Redhat thinks the ideas he has are so important that he can waste the time of everybody else.
With current cost of nuclear, it is not worthy an investment regardless how much fuel there is left.
Also I start to find the "if we would only invest another couple of hundred billions to develop a new generation of nuclear based on completely unproven designs (but trust us we never overpromised!) then it would be safe, burn the existing waste, and provide electricity to cheap to meter" bullshit the slashdot nuclear fanboys tend to fall for rather ridiculous. I have a bridge to sell you...
Offshore wind generated 2.7% of the power in Germany in 2017, but offshore wind is just starting now. All renewables were at 33.3%. Source: https://www.ag-energiebilanzen...
Electricity is expensive but only part of the price is due to the feed-in tariff for renewables. That this is paid for from the electricity price was intentional to reduce demand and avoid a rebound effect. It was also highly successful strategy as Germany is credited for bringing down the price for renewables. Nuclear also got a lot of subsidies from general taxes - this is not better. The energy transition in Germany is supported by large parts of the population, was discussed for decades, and well planned (with lots of research and large-scale simulations, e.g. by Fraunhofer which is a renowned engineering society). Also most nukes were not shut down directly, but based on life time - so old plants first. Coal use is lower than ever before. That Germany increased coal use is just a myth.
Coal use is a historical low in Germany (electricity production from coal 2017: 92.6 TWh lignite 2017: 147.5 TWh, vs. ten years ago: coal 2007: 142.0 TWh, lignite 2007: 155.1 WTh, source https://www.ag-energiebilanzen...).
Also the grid is pretty advanced and stable in the world while 33% of electricity is already produced by renewables and there is no indication of severe problems (certainly there are challenges, but no challenges which seem too hard to solve).
So coal is dropping and lignite is stable (there was some brief increase after Fukushima though). I agree it could be better, but the overall idea that Germany is using more coal is clearly nonsense.
It wasn't too useful as untrusted clients also can't use render and other important extensions. This also seems easily changed.
Not really. There's no accounting for buffer space, for example, or throttling for command streams in X.org. So one misbehaving client can still bring down everything. You can rewrite X.org completely from scratch using async primitives and threadsafe code everywhere. Sure. No problem at all.
Adding accounting for buffer space or throttling doesn't sound like major issues and I do not see why one would need to rewrite everything to add this. Not that this is the most pressing issue with modern desktop GUIs. I have not had a client bring down or hang X for a very very long time (and when this last happened years ago, this was a graphics driver issue). In contrast, those idiots rewriting everything have cost me hours or productive time by breaking the user interface of perfectly working tools again and again in the last decade.
Ok, so it starts from the other direction. So what? It isn't clear to me why - once all desired features and extensions have been added - the end result would be any better than what could be achieved by using isolated untrusted clients on X
The X.org codebase is crap. It can't be patched indefinitely, its core design is basically broken to be useful in a modern environment. The fact that somebody bothered to fake it with "security" extensions doesn't change it.
Again, the code base looks very reasonable to me. I have seen much worse. Please do not spread FUD. The core design is also not broken. In fact, the Wayland FAQ itself states: "It's entirely possible to incorporate the buffer exchange and update models that Wayland is built on into X" (and with present and DRI3 this has in fact already been done). So obviously, there is no fundamental problem with core X which would prevent such things as you incorrectly seem to believe.
So we now have to maintain both X and Wayland. Wow, Genius!
With XWayland you can have each application running its own private crappy version of X.org.
I can't find any real argument here. Your wording suggest that your judgement is a bit clouded by emotions. Having applications run their own copies of X will break useful communications mechanisms between different applications.
"Since many of the underlying technologies in these environments were not designed with strong application isolation in mind, users should only install applications using these interfaces from trusted sources." Solved?
Yep. Ubuntu snaps are as good as containment mechanisms in Linux (i.e. not that good at all).
Linux has excellent containment mechanisms. But the existence of this mechanisms does not imply that applications do not need to communicate or interact anymore, which needs careful consideration and design of safe APIs. I do not see how snaps solve this problem. Snaps also prevent centralized security fixes to libraries. This is the reason we abandoned static libraries a long time ago. This is not progress.
Why not? The necessary hooks seem to be all there. Untrusted X Clients are isolated against trusted X Clients. This seems to work.
Who told you this nonsense? X.org doesn't have any real distinction between clients. They are all similarly "trusted".
There's a thin veneer of SECURITY and XAce extensions which are not used by anybody. They are not even build-enabled in Debian and CentOS. They are also utterly inadequate, for example, SECURITY extension puts all "untrusted" clients together as there's no per-client isolation and doesn't prevent all sniffing.
Yes, that could easily be changed by also separating untrusted clients from each other. I once had a proof of concept patch doing exactly that and it was very simple... It wasn't too useful as untrusted clients also can't use render and other important extensions. This also seems easily changed.
A wayland shell would also be shared by its clients, or? With the security hooks already there, why is there are need for re-architecting?
Wayland clients don't have access to each other without shell explicitly granting capabilities to do it. They _might_ have access to the shell but it's easy to isolate. For example, a browser application can be limited only to submitting requests for direct rendering manager and to reading its input stream.
Ok, so it starts from the other direction. So what? It isn't clear to me why - once all desired features and extensions have been added - the end result would be any better than what could be achieved by using isolated untrusted clients on X.
I don't mind splitting up X or re-architecting parts of it (although I do not quite see why you think it is needed - we also have a monolithic kernel). The main problem I have with Wayland is that it breaks compatibility with the wire-protocol without a good reason.
Run XWayland for perfect backwards compat. Problem solved.
So we now have to maintain both X and Wayland. Wow, Genius!
Well, if they can interact, then isolation is difficult and somehow I doubt they have solved this completely, and if they can not really interact, they are useless....
"Since many of the underlying technologies in these environments were not designed with strong application isolation in mind, users should only install applications using these interfaces from trusted sources." Solved?
I pointed out the solution: X clients could be isolated against each other
No, they can not be isolated.
Why not? The necessary hooks seem to be all there. Untrusted X Clients are isolated against trusted X Clients. This seems to work.
X.org is shared across all clients and it can't be changed without re-architecting it.
A wayland shell would also be shared by its clients, or? With the security hooks already there, why is there are need for re-architecting?
Attempts to do it (XAce) died ignominiously.
True, but without proper access isolation for processes with the same UID, there was also no point.
You'll need to introduce multiple independent X servers and a layer above them to orchestrate input devices and access to direct rendering. In short, you'll get Wayland.
I don't mind splitting up X or re-architecting parts of it (although I do not quite see why you think it is needed - we also have a monolithic kernel). The main problem I have with Wayland is that it breaks compatibility with the wire-protocol without a good reason.
Wake me up if any Linux distribution ships with proper isolation between different programs from the same user.
Ding! Wake up call! Ubuntu does it quite well with snaps.
Well, if they can interact, then isolation is difficult and somehow I doubt they have solved this completely, and if they can not really interact, they are useless.... The other issue with Snap packages is that seem to come bundled with their own libraries. This seems to be a security nightmare.
I have to admit that I never looked at this part. It must be bad considering that this is where most prominent wayland developers also came from.
Except that in X there can be multiple applications at different privilege levels. That sudo in an X terminal? Totally vulnerable for an application that runs in a browser and has somehow gained access to the X socket.
I pointed out the solution: X clients could be isolated against each other. This would require some work, but the fundamental parts to make this work exist in X already for a long time.
Never mind that ptrace() can be effectively limited through a multitude of methods (LSM, namespaces, caps).
Wake me up if any Linux distribution ships with proper isolation between different programs from the same user. There are many issues and not just ptrace or X. And no, I do not see rewriting everything again and again in incompatible ways as an ideal strategy towards a solution.
You make it sound like it is a big deal, that there are rare days where Germany imports more electricity than it imports.
I'm sure it is rare. Yet a single grid destabilising power outage in even a portion of an economy the size of Germany is not a big deal. It's a HUGE fucking deal.
There was no such thing in Germany. In contrast, France had these kind of HUGE fucking deal issues. This is exactly my point.
At the same time, you fail to mention that France was often importing a significant amount of power continuously for weeks at a time in 2017 because it could not fulfill its own demand as too many nuclear plants were down.
You're comparing planned events to unplanned events. Don't do that. You may cause a power outage, and even one of the even a portion of an economy the size of France is not a big deal. It's a HUGE fucking deal.
I can't parse your ramblings. Nuclear plants have far more unplanned events than wind power which is actually well quite predictable.
Slashdot Mirror
It is not a terrible deep insight that wind is intermittent power source. It is also not new that there is less wind in the summer. On the other hand, there is more solar and the total production from renewables is surprisingly stable: https://www.energy-charts.de/r... Always about 35% which in summary is much more than nuclear.
Your charts highlight another point: The renewables clearly cut into consumption of coal and lignite. This is very good.
And there is another thing to learn: Nuclear is always running at the same level. This is related to a disadvantage of nuclear: It can *only* be used for baseload.Technically, it might be possible to do load following, but that makes the economics of nuclear even worse as you have very high investments you need to recover. On the one hand, this makes it very unattractive to have a lof of nuclear in a modern grid with intermittent power sources. On the other hand, you cannot use only nuclear because of demand side changes. So there is no real long-term use case for nuclear anymore.
How is a law preventing damage to the ecosystem of the rivers is a "political problem"?
I am not arguing against spending money to combat climate change, but the money we have available to combat climate change needs to be spent at cost effective solutions - and this means: not nuclear but wind, solar, storage, energy demand management, and more efficient technologies which reduce consumption. All these a much better use of the resources we have than subsidizing the nuclear industry which never delivered on their promises of cheap, safe, and reliable power.
Nuclear isn't the answer.
Then I'd like to hear what is the answer. Think quickly because the clock is ticking.
The answers are renewables, storage, energy demand management, and saving.
It was promised to be too cheap to meter, instead it is the most costly to generate.
That's kind of irrelevant now, no?
No it is not. In a world of limited resources, you can not afford to waste money on an expensive technology as it reduces the money you can spend on more efficient solutions.
That might have something to do with those unwashed hippies that have been trying to "save the planet". The planet's fine, it's us humans that are fucked if we don't do something.
Unwashed hippies don't have much political influence. So no.
The only plant currently being built (Georgia) is 5 years behind schedule and double its original cost.
Then throw some more money at it. The alternative is potential extinction.
The alternative is to spend to money on more reasonable solutions.
This is electricity generation in Gemany from renewables from 1900 to 2017 in TWh / year:
19,7 25,1 37,9 38,9 46,1 46,1 57,2 63,1 72,4 89,1 94,1 95,7 105,2 123,6 143,3 152,5 162,5 188,6 189,8 218,3
This is substantial amount of power. At the same time price for renewables dropped, wind is basically competitive (hydro has been for long time) and for photovoltaics it is just a matter of time as price drop exponentially: https://en.wikipedia.org/wiki/...
In contrast, the huge amount of investments in nuclear which were ongoing for more than half a center did not achieve something remotely similar. It is still an inefficient technology which cannot compete without subsidies.
This is nonsense.
This is power generation in Germany from coal and lignite from 1990-2018 in TWh.
coal 140,8 147,1 143,1 138,4 134,6 146,5 140,8 134,1 137,9 142,0 124,6 107,9 117,0 112,4 116,4 127,3 118,6 117,7 112,2 92,6
lignite 170,9 142,6 148,3 154,8 158,0 158,2 158,0 154,1 151,1 155,1 150,6 145,6 145,9 150,1 160,7 160,9 155,8 154,5 149,5 147,5
Also the CO2 emission from electricity production decreased from 315 Mio t CO2 emission in 2010 (before shutting down a couple of nukes in response to Fukishima) to 285 Mio t CO2 in 2017. At the same time power production increased from 564 TWh to 583 TWh.
Well, you stated that CO2 emissions increased in Germany since shutting down nuclear power plants. Technically this may be true, but this is highly misleading as CO2 emissions from electricity generation in Germany actually *decreased* and the overall increase (it it indeed exists for this time frame, I haven't checked) from CO2 comes from elsewhere (likely transportation). The CO2 emission from electricity production are 315 Mio t CO2 emission in 2010 (before shutting down a couple of nukes in response to Fukishima) to 285 Mio t CO2 in 2017. At the same time power production increased from 564 TWh to 583 TWh. So Germany decreased CO2 emissions from electricity production even after shutting down several nukes and increasing output. Whatever argument you tried to make, it based on a wrong premise.
The US is one of the countries with the highest per capita CO2 emissions per year substantially higher than Germany, Japan, China (and a lot higher than India.) This while having a trade deficit. No, we do not expect to pay for all other nations, we expect that you bring down *your* emissions.
That Germany is adding coal plants is a myth.
This is power generation from coal and lignite in Germany from 1990-2017 in TWh.
coal 140,8 147,1 143,1 138,4 134,6 146,5 140,8 134,1 137,9 142,0 124,6 107,9 117,0 112,4 116,4 127,3 118,6 117,7 112,2 92,6
lignite 170,9 142,6 148,3 154,8 158,0 158,2 158,0 154,1 151,1 155,1 150,6 145,6 145,9 150,1 160,7 160,9 155,8 154,5 149,5 147,5
Source: https://www.ag-energiebilanzen...
Correction: 1990-2017
Electricity generation has steadily reduced CO2 emissions in Germany. This has been offset by an increase in other sectors (transportation mostly). Making an argument against wind/solar is completely misleading.
This is power generation from coal and lignite from 1990-2018 in TWh.
coal 140,8 147,1 143,1 138,4 134,6 146,5 140,8 134,1 137,9 142,0 124,6 107,9 117,0 112,4 116,4 127,3 118,6 117,7 112,2 92,6
lignite 170,9 142,6 148,3 154,8 158,0 158,2 158,0 154,1 151,1 155,1 150,6 145,6 145,9 150,1 160,7 160,9 155,8 154,5 149,5 147,5
Source: https://www.ag-energiebilanzen...
If I have to first go to a mailing list of my distribution to learn about some random "reasons" to why some change is supposed to be good than this is almost by definition "no good reason". Is an actual improvement noticeable to the user too much to ask for?
I am scientist. I have to learn new stuff every day. I develop new stuff every day. But I have no sympathy for people wasting my time by breaking standard tools or conventions with no good reason. And the "you are just to lazy to learn new things" argument is just BS. I want to spend my time learning interesting things and not have to relearn how to do basic stuff with my computer because some random dude at Redhat thinks the ideas he has are so important that he can waste the time of everybody else.
The dams which might kill millions when they burst are usually dams for flood control ... not power plants.
With current cost of nuclear, it is not worthy an investment regardless how much fuel there is left.
Also I start to find the "if we would only invest another couple of hundred billions to develop a new generation of nuclear based on completely unproven designs (but trust us we never overpromised!) then it would be safe, burn the existing waste, and provide electricity to cheap to meter" bullshit the slashdot nuclear fanboys tend to fall for rather ridiculous. I have a bridge to sell you...
offshore wind produce 2.7% of all electricity in Germany in 2017. Source: https://www.ag-energiebilanzen...
Offshore wind generated 2.7% of the power in Germany in 2017, but offshore wind is just starting now. All renewables were at 33.3%. Source: https://www.ag-energiebilanzen...
Electricity is expensive but only part of the price is due to the feed-in tariff for renewables. That this is paid for from the electricity price was intentional to reduce demand and avoid a rebound effect. It was also highly successful strategy as Germany is credited for bringing down the price for renewables. Nuclear also got a lot of subsidies from general taxes - this is not better. The energy transition in Germany is supported by large parts of the population, was discussed for decades, and well planned (with lots of research and large-scale simulations, e.g. by Fraunhofer which is a renowned engineering society). Also most nukes were not shut down directly, but based on life time - so old plants first. Coal use is lower than ever before. That Germany increased coal use is just a myth.
Coal use is a historical low in Germany (electricity production from coal 2017: 92.6 TWh lignite 2017: 147.5 TWh, vs. ten years ago: coal 2007: 142.0 TWh, lignite 2007: 155.1 WTh, source https://www.ag-energiebilanzen...).
Also the grid is pretty advanced and stable in the world while 33% of electricity is already produced by renewables and there is no indication of severe problems (certainly there are challenges, but no challenges which seem too hard to solve).
It helps to look at actual numbers (energy production in TWh / per year):
coal: 143.1 (2000) -> 117.0 (2010) -> 92.6 (2017)
lignite: 148.3 (2000) -> 145.9 (2010) -> 147.5 (2017)
So coal is dropping and lignite is stable (there was some brief increase after Fukushima though). I agree it could be better, but the overall idea that Germany is using more coal is clearly nonsense.
Source: https://www.ag-energiebilanzen...
Looks decreasing to me:
https://knoema.com/atlas/Germa...
https://knoema.com/atlas/Unite...
It wasn't too useful as untrusted clients also can't use render and other important extensions. This also seems easily changed.
Not really. There's no accounting for buffer space, for example, or throttling for command streams in X.org. So one misbehaving client can still bring down everything. You can rewrite X.org completely from scratch using async primitives and threadsafe code everywhere. Sure. No problem at all.
Adding accounting for buffer space or throttling doesn't sound like major issues and I do not see why one would need to rewrite everything to add this. Not that this is the most pressing issue with modern desktop GUIs. I have not had a client bring down or hang X for a very very long time (and when this last happened years ago, this was a graphics driver issue). In contrast, those idiots rewriting everything have cost me hours or productive time by breaking the user interface of perfectly working tools again and again in the last decade.
Ok, so it starts from the other direction. So what? It isn't clear to me why - once all desired features and extensions have been added - the end result would be any better than what could be achieved by using isolated untrusted clients on X
The X.org codebase is crap. It can't be patched indefinitely, its core design is basically broken to be useful in a modern environment. The fact that somebody bothered to fake it with "security" extensions doesn't change it.
Again, the code base looks very reasonable to me. I have seen much worse. Please do not spread FUD. The core design is also not broken. In fact, the Wayland FAQ itself states: "It's entirely possible to incorporate the buffer exchange and update models that Wayland is built on into X" (and with present and DRI3 this has in fact already been done). So obviously, there is no fundamental problem with core X which would prevent such things as you incorrectly seem to believe.
So we now have to maintain both X and Wayland. Wow, Genius!
With XWayland you can have each application running its own private crappy version of X.org.
I can't find any real argument here. Your wording suggest that your judgement is a bit clouded by emotions. Having applications run their own copies of X will break useful communications mechanisms between different applications.
"Since many of the underlying technologies in these environments were not designed with strong application isolation in mind, users should only install applications using these interfaces from trusted sources." Solved?
Yep. Ubuntu snaps are as good as containment mechanisms in Linux (i.e. not that good at all).
Linux has excellent containment mechanisms. But the existence of this mechanisms does not imply that applications do not need to communicate or interact anymore, which needs careful consideration and design of safe APIs. I do not see how snaps solve this problem. Snaps also prevent centralized security fixes to libraries. This is the reason we abandoned static libraries a long time ago. This is not progress.
Why not? The necessary hooks seem to be all there. Untrusted X Clients are isolated against trusted X Clients. This seems to work.
Who told you this nonsense? X.org doesn't have any real distinction between clients. They are all similarly "trusted".
There's a thin veneer of SECURITY and XAce extensions which are not used by anybody. They are not even build-enabled in Debian and CentOS. They are also utterly inadequate, for example, SECURITY extension puts all "untrusted" clients together as there's no per-client isolation and doesn't prevent all sniffing.
Yes, that could easily be changed by also separating untrusted clients from each other. I once had a proof of concept patch doing exactly that and it was very simple... It wasn't too useful as untrusted clients also can't use render and other important extensions. This also seems easily changed.
A wayland shell would also be shared by its clients, or? With the security hooks already there, why is there are need for re-architecting?
Wayland clients don't have access to each other without shell explicitly granting capabilities to do it. They _might_ have access to the shell but it's easy to isolate. For example, a browser application can be limited only to submitting requests for direct rendering manager and to reading its input stream.
Ok, so it starts from the other direction. So what? It isn't clear to me why - once all desired features and extensions have been added - the end result would be any better than what could be achieved by using isolated untrusted clients on X.
I don't mind splitting up X or re-architecting parts of it (although I do not quite see why you think it is needed - we also have a monolithic kernel). The main problem I have with Wayland is that it breaks compatibility with the wire-protocol without a good reason.
Run XWayland for perfect backwards compat. Problem solved.
So we now have to maintain both X and Wayland. Wow, Genius!
Well, if they can interact, then isolation is difficult and somehow I doubt they have solved this completely, and if they can not really interact, they are useless....
Solved: https://github.com/snapcore/sn...
"Since many of the underlying technologies in these environments were not designed with strong application isolation in mind, users should only install applications using these interfaces from trusted sources." Solved?
I pointed out the solution: X clients could be isolated against each other
No, they can not be isolated.
Why not? The necessary hooks seem to be all there. Untrusted X Clients are isolated against trusted X Clients. This seems to work.
X.org is shared across all clients and it can't be changed without re-architecting it.
A wayland shell would also be shared by its clients, or? With the security hooks already there, why is there are need for re-architecting?
Attempts to do it (XAce) died ignominiously.
True, but without proper access isolation for processes with the same UID, there was also no point.
You'll need to introduce multiple independent X servers and a layer above them to orchestrate input devices and access to direct rendering. In short, you'll get Wayland.
I don't mind splitting up X or re-architecting parts of it (although I do not quite see why you think it is needed - we also have a monolithic kernel). The main problem I have with Wayland is that it breaks compatibility with the wire-protocol without a good reason.
Wake me up if any Linux distribution ships with proper isolation between different programs from the same user.
Ding! Wake up call! Ubuntu does it quite well with snaps.
Well, if they can interact, then isolation is difficult and somehow I doubt they have solved this completely, and if they can not really interact, they are useless.... The other issue with Snap packages is that seem to come bundled with their own libraries. This seems to be a security nightmare.
I worked on the input stack.
I have to admit that I never looked at this part. It must be bad considering that this is where most prominent wayland developers also came from.
Except that in X there can be multiple applications at different privilege levels. That sudo in an X terminal? Totally vulnerable for an application that runs in a browser and has somehow gained access to the X socket.
I pointed out the solution: X clients could be isolated against each other. This would require some work, but the fundamental parts to make this work exist in X already for a long time.
Never mind that ptrace() can be effectively limited through a multitude of methods (LSM, namespaces, caps).
Wake me up if any Linux distribution ships with proper isolation between different programs from the same user. There are many issues and not just ptrace or X. And no, I do not see rewriting everything again and again in incompatible ways as an ideal strategy towards a solution.
You make it sound like it is a big deal, that there are rare days where Germany imports more electricity than it imports.
I'm sure it is rare. Yet a single grid destabilising power outage in even a portion of an economy the size of Germany is not a big deal. It's a HUGE fucking deal.
There was no such thing in Germany. In contrast, France had these kind of HUGE fucking deal issues. This is exactly my point.
At the same time, you fail to mention that France was often importing a significant amount of power continuously for weeks at a time in 2017 because it could not fulfill its own demand as too many nuclear plants were down.
You're comparing planned events to unplanned events. Don't do that. You may cause a power outage, and even one of the even a portion of an economy the size of France is not a big deal. It's a HUGE fucking deal.
I can't parse your ramblings. Nuclear plants have far more unplanned events than wind power which is actually well quite predictable.