Conversation
Jarkko Sakkinen
jarkko
Is this the first rust resistance type of article of the year: https://lwn.net/Articles/1006805/
I thought we got over it and now it is just yet another boring language we use (which is great) :D
I thought we got over it and now it is just yet another boring language we use (which is great) :D
1
0
0
@jarkko I thought so too 😅
BTW are you doing the new version v4l2 loopback in Rust ? 😶🌫️🙈
1
0
1
Jarkko Sakkinen
jarkko
@Aissen
I did that consideration actually.
Here's the summary of reasons of using C for that driver at minimum for the first RFC patch set:
1. The existing code base is in C.
2. I'm not an expert on V4L2.
3. I'm pretty seasoned with the Linux driver model.
I.e. if you start to do too much to the domain per, there might be something "unimportant looking" that you look as "garbage", while you're actually removing something that does something useful.
This is IMHO pretty sound way to look at it. Since it is my hobby project given feedback, I might then re-consider re-implementing it in Rust. It is actually good comment to add to the cover letter.
IMHO, no matter who was doing this I make the right *first* call both from "respect the creators" and also technical angle. I cycled SGX patch set for over two years so writing the same driver a few times is like nothing ;-)
I did that consideration actually.
Here's the summary of reasons of using C for that driver at minimum for the first RFC patch set:
1. The existing code base is in C.
2. I'm not an expert on V4L2.
3. I'm pretty seasoned with the Linux driver model.
I.e. if you start to do too much to the domain per, there might be something "unimportant looking" that you look as "garbage", while you're actually removing something that does something useful.
This is IMHO pretty sound way to look at it. Since it is my hobby project given feedback, I might then re-consider re-implementing it in Rust. It is actually good comment to add to the cover letter.
IMHO, no matter who was doing this I make the right *first* call both from "respect the creators" and also technical angle. I cycled SGX patch set for over two years so writing the same driver a few times is like nothing ;-)
1
0
1
@Aissen Thanks for asking that. I'll put a remark to the cover letter. It would not be a problem for me personally. My argument comes from more like doing right things right from the process angle
Unfortunately Ihave had not time to push forward since I've been fixing Windows build issue :-) After I've done my release critical fixes for the next Polkadot release (for PolkaVM) I plan to put focus on this again.
Fixing Windows build was good learning otho. This is how you bootstrap a build for arbitrary target so that the build system won't trip and it will work out Linux/macOS/Windows:
https://github.com/paritytech/polkavm/blob/7715ad466e2d084e86e01e8341eb53ad0d86c3c0/crates/polkavm-test-data/build.rs
Nothing complicated but finding the simple recipe that actually works is sometimes tedious :-) I did not find any example that would be as generically applicable as this...
Unfortunately Ihave had not time to push forward since I've been fixing Windows build issue :-) After I've done my release critical fixes for the next Polkadot release (for PolkaVM) I plan to put focus on this again.
Fixing Windows build was good learning otho. This is how you bootstrap a build for arbitrary target so that the build system won't trip and it will work out Linux/macOS/Windows:
https://github.com/paritytech/polkavm/blob/7715ad466e2d084e86e01e8341eb53ad0d86c3c0/crates/polkavm-test-data/build.rs
Nothing complicated but finding the simple recipe that actually works is sometimes tedious :-) I did not find any example that would be as generically applicable as this...
2
0
1
@jarkko I had issues with windows builds as well. Much easier to resolve than what you did! https://social.treehouse.systems/@Aissen/112627256950370265
1
0
1