tokio
clap
iced
error_chain (and the list of newer and shinier error handling crates)
serde
serde_derive
serde_json
yaml
log
env_logger
url
tempdir
toml
Is a modified recipe.toml required in order to compile this branch/MR? If so, could you drop a copy here. Thanks
I checked out the branch, but I still see the old travis CI files, and not the newer gitlab versions - so it makes me think it has not been rebased on upstream/master?
That's an excellent small reproducer that should help track it down.
I have reworking of pkgutils and hack in tokio to get pkgutils compiling on aarch64. I plan to look at installer next to see if I can get that compiling on aarch64 also.
My MR that was merged to drivers-initfs should fix that part workaround of yours. Can you remove your patch, rebuild and check?
Maybe list the things you had to modify, skip, workaround in order to get this far (e.h. pkgutils, etc), so we can generate other Issues to try and fix them also? (or create the issues directly is another option i guess).
Great!
Andrew Mackenzie (b004a667) at 06 May 18:37
Andrew Mackenzie (b004a667) at 06 May 18:36
Correct the list of drivers to include
Only build the required drivers for each architecture
Andrew Mackenzie (9b0b0b34) at 06 May 17:53
Update to newer version of redoxer and allow patch updates
... and 1 more commit
Do you need backend-winit-x11 even if you are Ubuntu with Wayland? (I am not sure what distro is used for "linux" in CI).
Do you want to take a shot at an MR with those features, conditional on the target_os in Cargo.toml?
Then, I could confirm on my linux (Ubuntu) and we can try adding calculator to CI (on linux) and close out this issue!
I see it's a strange workspace, split into two (I assume to enable starting the slint ports), where one member in the calculator slint port (would be better to be "slint" with all slint ports under it) and "orbutils" that is a single member with a bunch of binĂ ries.
One advantage of a workspace is really to reduce compile time by sharing the same dependencies that get compiled once (not per project) into a shared target dir....
Since there is no lib, or shared code between the binaries....they could just as easily be completely separate crates....just all in the same repo (so a Cargo.toml file in each subdir for each binary). The disadvantage of that is when you have to compile/check/test/publish all of them....you need a script to do the cargo command for each one (instead of doing all with one cargo command in the workspace root.
Another direction would be a separate repo, crate and recipe for each "util" (like orbterm is separate now) as the list grows, not forcing them all to be together.
Andrew Mackenzie (befe3880) at 02 May 22:13
Arguably the parser in package.rs should be robust to this, but this is the simpler fix for now.
Andrew Mackenzie (befe3880) at 02 May 21:38
remove blank line
Andrew Mackenzie (5c0c5a2c) at 02 May 16:19
Removed it
Andrew Mackenzie (5c0c5a2c) at 02 May 15:23
Remove feature