-
Notifications
You must be signed in to change notification settings - Fork 78
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Tweak Rust package guidelines #670
Comments
Unfortunately, neither |
Bummer. Also a bummer this never got flushed out. Also a bummer my patch fixing the So why is Rust coming up with a different architecture name? Is this something it's impossible to normalize? |
Hi @alerque , I think that's partially because the compiler need to construct the |
I think that there can be two possible solutions. One is submitting patch to Rust and change the naming scheme to something like arch= |
Bah. So basically right now the only way to solve both my concern and eliminate the need for downstream patching would be something like this: cargo fetch --locked --target "$(rustc -vV | awk '/^host:/ { print $2 }')" Obviously that's a bit convoluted so I'm not recommending it yet, just trying to get my head around the problem here. It seems to me the best course of action would be to actually fix
As far as I can see there is no way to actually use this so called "default" in fetch, it only defaults to "all architectures" (which for some dependency chains is becoming pretty heavy!). |
Hi @alerque, I recreated your PR for cargo's doc change as rust-lang/cargo#10398 👀 |
This patch fix three error: 1. The [target specification issue]. 2. Crate libc can not be compiled due to symbol missing. This is fixed in [rust-lang/libc PR felixonmars#2668], and release in 0.2.118 version. 3. Test failure Some tests of diskonaut depend on the previous events. It take a snapshot of the TUI pixels output, then compare those snapshot pixel by pixel to determine if the cli is working correctly. However those test are not reliable. If the previous event took too long to finish, the TUI output will not match what is expected. The current workaround is to increase the event interval. One second interval is enough for all event to finish its jobs. Tested in 5950x QEMU environment. Ref: * [target specification issue]: felixonmars#670 (comment) * [rust-lang/libc PR felixonmars#2668]: rust-lang/libc#2668 Signed-off-by: Avimitin <[email protected]>
This patch fix three error: 1. The [target specification issue]. 2. Crate libc can not be compiled due to symbol missing. This is fixed in [rust-lang/libc PR #2668], and release in 0.2.118 version. 3. Test failure Some tests of diskonaut depend on the previous events. It take a snapshot of the TUI pixels output, then compare those snapshot pixel by pixel to determine if the cli is working correctly. However those test are not reliable. If the previous event took too long to finish, the TUI output will not match what is expected. The current workaround is to increase the event interval. One second interval is enough for all event to finish its jobs. Tested in 5950x QEMU environment. Ref: * [target specification issue]: #670 (comment) * [rust-lang/libc PR #2668]: rust-lang/libc#2668 Signed-off-by: Avimitin <[email protected]>
Closing since there has been a working Rust package guidelines for a while. :) |
I don't have any RISC boards, but poking around in here one thing that seems to need patching a lot is something I put in the Rust packaging guidelines.
https://github.com/felixonmars/archriscv-packages/blob/82d8d619e862478a076ae657069c9caf2f1caed4/onefetch/riscv64.patch
This platform limited fetch can mean a huge savings in downloads when building packages (in some cases from hundreds of MB down to less than 10), so I am loath to drop it altogether.
Would it be helpful to modify the Rust recommendations like so:
...or is the issue here that you are cross compiling and need to fetch far an architecture other than the host? If so is there a variable we could suggest people use to get the benefit of only downloading required sources while not making so much work patching downstream?
The text was updated successfully, but these errors were encountered: