Microsoft Joins Open Invention Network With Its 60,000+ Patents

[stargazer]

I've heard stories for years that MS doesn't make nearly as much money for it's OS as it used to, and that was why it focused more on it's Office suite. I've also heard stories that the Window OS is a spaghetti mess that would cost MS a fortune to fully clean up. When Apple re-based MacOS on BSD, I figured it was inevitable that MS would eventually re-base on an open source platform as well. When MS released Windows Subsytem for Linux, I figured that was another piece of the puzzle for their eventual move. This patent release is another piece of the puzzle. Ultimately MS pays a huge amount of money developing their own OS core that they could eliminate if they built their OS around an open source platform. I think we will see it in our lifetimes, it's just going to take a long time for MS to work around their own legacy baggage to make it happen. In the interim, you'll see baby steps like this.

[duby229]

And has quite different kernel design from Windows too. Not everything can be matched 1:1.

Yeah, differences like Linux has a real process scheduler and can account for multi-core and SMT. Linux has a real actual memory manager and and it only swaps to storage in the last case scenario.... I can give at least another ten more reasons why Linux has massive performance advantages over windows. And you can bet your last dollar it will translate to Vulkan and hence also DXVK. I wouldn't be at all surprised if Directx on Linux seriously outperforms Windows some day.

EDIT: In many different ways, windows bottlenecks on storage. It is totally incapable of -not- bottlenecking on storage. It starts at IO, which doesn't have a real scheduler. It continues on with swap, which is constant. And it's made far worse because things like search indexing or superfetch are constantly thrashing storage which makes available bandwidth even worse for things like swapping....

There is no chance in any hell that windows could outperform linux ever. If you don't believe me just boot up windows and look at your cases activity LED.......

[schmidtbag]

I find the timing of this quite interesting. With Linus temporarily not being the primary maintainer, MS could have seen this opportunity as "now's our chance to sneak in". I don't know exactly what MS could possibly do, but they're not known for their generosity. Keep in mind a single patent could cost several thousand dollars (including lawyer fees, paying those who wrote them up, the engineers who designed the feature, etc). MS must be getting something out of this.

[ZeroPointEnergy]

It's different. Doesn't Microsoft sell cloud Linux servers now?

They embrace Linux on the server now because it becomes the default OS in the cloud and they don't want to miss out on that revenue stream. That is the only thing that is different. All their products are still proprietary and closed source. There is nothing visible of that open source love on the desktop side of things where they don't have a direct profit from it. This is not enough to earn the trust and make everyone forget about the past, not by a long shot.

[Bigon]

That'd be awesome, they've been doing all that linux interopability for Windows 10, would be great if they do some of that the other way around.

One thing that's bugged me with USB storage(at least FAT32, not sure if it affects others) on linux is that common issue people experience where you write data and a DE reports it as a completed write/transfer, but in reality, the kernel has just written the data to write to memory and still flushing it to the actual storage media. Or you get what appears to be a normal write(or the write speed is much higher than it should be because it's actually writing to memory), and then it slows down to snail pace and can take several hours for a GB or so to actually finish writing to the storage media, and only then can you unmount it. If you unmount prior, the write wasn't completed and your data is corrupt...

IIRC, I found out this was due to the kernels vm (virtual memory?) parameters, which while suitable for internal media, had buffers and other things too large for external media(might actually be from USB 2.0 devices, but I know the perf is a snail compared to Windows). You could adjust these but not per device, it was unfortunately global, so your internal media would take a dive in I/O perf, unless you specifically adjusted it momentarily for your external media write.

The disk could be mounted with the "sync" option, that would mean no caching at all

[duby229]

I find the timing of this quite interesting. With Linus temporarily not being the primary maintainer, MS could have seen this opportunity as "now's our chance to sneak in". I don't know exactly what MS could possibly do, but they're not known for their generosity. Keep in mind a single patent could cost several thousand dollars (including lawyer fees, paying those who wrote them up, the engineers who designed the feature, etc). MS must be getting something out of this.

They think they are going to get a much, much better kernel..... EEE.....

[caligula]

One thing that's bugged me with USB storage(at least FAT32, not sure if it affects others) on linux is that common issue people experience where you write data and a DE reports it as a completed write/transfer, but in reality, the kernel has just written the data to write to memory and still flushing it to the actual storage media. Or you get what appears to be a normal write(or the write speed is much higher than it should be because it's actually writing to memory), and then it slows down to snail pace and can take several hours for a GB or so to actually finish writing to the storage media, and only then can you unmount it. If you unmount prior, the write wasn't completed and your data is corrupt...

Guess what, your USB media has some shitty class 2, 1/2 Mbps flash with a crappy 10 IOPS controller. I also used to run into these issues, but after switching to good brand USB3 sticks and professional SD cards, the issues are all gone. Besides, the aggressive buffering you are complaining here actually improves the write performance as the disk can we written with as few IOPS as possible. In the GUI, just press the 'eject' button in the menus and wait till you get the notification regarding finished writes. It works just fine. Or use MTP/PTP protocols instead of USB mass storage.

[caligula]

One case was copying over an Ubuntu iso onto USB instead of just using dd to create a boot image then and there, can't recall why I was transferring the iso to USB, but I remember it taking several hours to complete, another one was a 2-3GB file that I think took 7 hours+(different machine).

Stop using those shitty el cheapo usb sticks. You really get what you paid for. I've burned thousands of images for various purposes on several machines (from RPi 2 to 16 core Xeons) and it more or less boils down to buggy USB mass storage / UAS firmware or cheap, slow, low quality flash drives that choke on data. Nowadays my good quality sticks and SD cards can store 100+ MegaBytes/s. On USB2 only 35 MegaBytes/s, but that's still ok.

[linuxUser]

I'm surprised to see this Microsoft propaganda published here without telling the truth! Microsoft does most of it's patent attacks now through patent trolls. The OIN does not stop this. Microsoft is polishing its image here, no change is taking place in there bad behavior. How many times will people fall for the same bullshit stories? Nothing has changed here.

"Bruce Perens said not too long ago, OIN exists to protect software patents from the community rather than protect the community from software patents."

Read the article in the link below and educate yourself, Microsoft is not your friend.

[ms178]

I've heard stories for years that MS doesn't make nearly as much money for it's OS as it used to, and that was why it focused more on it's Office suite. I've also heard stories that the Window OS is a spaghetti mess that would cost MS a fortune to fully clean up. When Apple re-based MacOS on BSD, I figured it was inevitable that MS would eventually re-base on an open source platform as well. When MS released Windows Subsytem for Linux, I figured that was another piece of the puzzle for their eventual move. This patent release is another piece of the puzzle. Ultimately MS pays a huge amount of money developing their own OS core that they could eliminate if they built their OS around an open source platform. I think we will see it in our lifetimes, it's just going to take a long time for MS to work around their own legacy baggage to make it happen. In the interim, you'll see baby steps like this.

I thought they were busy with their CoreOS approach? Did they bury it recently alongside the whole UWP strategy? Opening up for recent industry wide efforts, like LLVM/Clang, Vulkan, the Linux Kernel and ecosystem would be really great to see. And I guess sharing all of the development burden with the rest of the industry while making money with other software products and services could make business sense for them. But I guess we are not there yet.