[0dayz]

>systemd

When will this tired piece of shit known as FUD die.

With that logic then the damn Linux kernel got "unexpected" ties to Microsoft.

>The trend is that its becoming more and more difficult (due to reimplementation costs and dependency lock-ins) to do Linux in the non-Microsoft way.

No idea what this is suppose to mean in this context, that we get new hardware stanards/tech like UEFI/GPT/secure boot or that people choose to use software like systemd?

[blueflow]

That these standards are written with the interests of Microsoft in mind. And that the only Linux-side implementation of them is systemd, where an Microsoft employee is the head maintainer. So if you want any of the new "features" you will need to have a linux/systemd/GPT/UEFI stack.

Smells alot like embrace, extend. Extinguish comes next.

[0dayz]

All I see from your comment: With a hammer in hand, everything is a nail.

Or to be more specific: If Microsoft is directly or directly involved in OSS or standards, then it's all EEE/compromised.

GPT/UEFI are standards, they are not ruled by Microsoft nor do I even think it was sole written with Microsoft in mind, at best what I can find is that UEFI took the same architecture Windows has with EFI, that's about it.

systemd's only concern with well systemd, no one is stopping you from not using it, or I'm oh so curious how non-systemd distros are able to boot WITH UEFI + GPT.

[blueflow]

> GPT/UEFI are standards, they are not ruled by Microsoft

- GPT partition uuid's with the "mixed endianess" from Windows COM: https://devblogs.microsoft.com/oldnewthing/20220928-00/?p=10...

- ESP is the FAT filesystem, developed in 1977 for DOS

- LFS vfat extensions so the ESP can have directory entries with lowercase letters or longer than 11 characters. This also pulls in the UCS-2 encoding. Stuff from Windows 95.

- EFI executables are actually windows PE executables

- ... which are hacked on top of a MS-DOS MZ executables (The "This program cannot be run in DOS mode" thing)

- ... and use the Windows64 calling convention

So formally, UEFI is not ruled by Microsoft. Somehow yet all this Windows/MS-DOS stuff ended up in it.

[0dayz]

>GPT partition uuid's with the "mixed endianess" from Windows COM

From the wikipedia[1]:

>The binary encoding of UUIDs varies between systems. Variant 1 UUIDs, nowadays the most common variant, are encoded in a big-endian format.

>Variant 2 UUIDs, historically used in Microsoft's COM/OLE libraries, use a little-endian format, but appear mixed-endian with the first three components of the UUID as little-endian and last two big-endian, due to the missing byte dashes when formatted as a string

In other words, there are 2 variants, 2 is for Microsoft's COM/OLE, so it's not designed by Microsoft.

>ESP is the FAT filesystem, developed in 1977 for DOS

>LFS vfat extensions so the ESP can have directory entries with lowercase letters or longer than 11 characters. This also pulls in the UCS-2 encoding. Stuff from Windows 95.

FAT is not govern by Microsoft, despite them being the original author, and even then FAT32 patents have expired[2].

>EFI executables are actually windows PE executables

You would need to provide proper source for this, the only source I can only talks about boot loaders, which I would imagine is written with the OS in mind[3].

[1]https://en.wikipedia.org/wiki/Universally_unique_identifier#...

[2]https://en.wikipedia.org/wiki/File_Allocation_Table#Patents

[3]https://en.wikipedia.org/wiki/UEFI#Applications

[blueflow]

You are wrong on the UUIDs, Variant 2 is what Microsoft did and got standardized after the fact.

And here the history and current usage of the PE format: https://en.wikipedia.org/wiki/Portable_Executable

It was made for Windows 3.1.

[0dayz]

If you read my comment or the Wikipedia page you'll see clear as day that varient 2 is meant for windows, so I dont get this correction.

And again that source doesn't give me any evidence of your claim.

[pjmlp]

First of all Linux distributions are the laggard, commercial UNIX systems have moved away from init before systemd was an idea.

Secondly, Lennart Poettering became a Microsoft employee only recently, after all the systemd drama took place.

[blueflow]

launchd does none of the whacky bootloader stuff that systemd does.

And Poettering didn't stop working on systemd, so you can estimate that Microsoft is not having a negative attitude towards his previous work.

[pjmlp]

launchd is one among many.

systemd adoption across Linux distributions and the features everyone complains about, predates his hiring process by several years.

[blueflow]

Which other "init system" is mucking with the bootloader and partition table like systemd does? How is that even in-scrope for an init system?

[pjmlp]

Solaris SMF for example,

https://docs.oracle.com/cd/E36784_01/html/E36801/svcrun-123....

https://docs.oracle.com/cd/E23824_01/html/821-2731/ggnke.htm...