Closed rbradford closed 2 years ago
I had to update the PR since there were other uses of vm-memory
in this repository. But I think you have a fundamental problem with the virtio-vsock
crate as that has a strict dependency on virtio-queue
0.2.0 which will be pulled from crates.io with the fixed vm-memory
version.
I had to update the PR since there were other uses of
vm-memory
in this repository. But I think you have a fundamental problem with thevirtio-vsock
crate as that has a strict dependency onvirtio-queue
0.2.0 which will be pulled from crates.io with the fixedvm-memory
version.
Thanks for the heads up, we will probably have a release of virtio-queue sooner than one for the vsock, and we can update then the virtio-queue dependency with 0.3.0.
I had to update the PR since there were other uses of
vm-memory
in this repository. But I think you have a fundamental problem with thevirtio-vsock
crate as that has a strict dependency onvirtio-queue
0.2.0 which will be pulled from crates.io with the fixedvm-memory
version.Thanks for the heads up, we will probably have a release of virtio-queue sooner than one for the vsock, and we can update then the virtio-queue dependency with 0.3.0.
The CI is failing because of this very problem because something is trying to link to two the two versions. I think this probably needs manual intervention of making a new release of the virtio-queue crate with the relaxed dependencies.
I had to update the PR since there were other uses of
vm-memory
in this repository. But I think you have a fundamental problem with thevirtio-vsock
crate as that has a strict dependency onvirtio-queue
0.2.0 which will be pulled from crates.io with the fixedvm-memory
version.Thanks for the heads up, we will probably have a release of virtio-queue sooner than one for the vsock, and we can update then the virtio-queue dependency with 0.3.0.
The CI is failing because of this very problem because something is trying to link to two the two versions. I think this probably needs manual intervention of making a new release of the virtio-queue crate with the relaxed dependencies.
One solution would be to make the virtio-vsock
crate use the virtio-queue
crate in the repo instead. This is how the virtio-blk
crate does it.
I had to update the PR since there were other uses of
vm-memory
in this repository. But I think you have a fundamental problem with thevirtio-vsock
crate as that has a strict dependency onvirtio-queue
0.2.0 which will be pulled from crates.io with the fixedvm-memory
version.Thanks for the heads up, we will probably have a release of virtio-queue sooner than one for the vsock, and we can update then the virtio-queue dependency with 0.3.0.
The CI is failing because of this very problem because something is trying to link to two the two versions. I think this probably needs manual intervention of making a new release of the virtio-queue crate with the relaxed dependencies.
One solution would be to make the
virtio-vsock
crate use thevirtio-queue
crate in the repo instead. This is how thevirtio-blk
crate does it.
Yes, we can also use path
together with version
. But it's a bit strange, the crates should be as independent as possible, i.e. virtio-vsock build with virtio-queue 0.2.0, and the others with the current HEAD.
One solution would be to make the
virtio-vsock
crate use thevirtio-queue
crate in the repo instead. This is how thevirtio-blk
crate does it.
My PR now has a patch that does this. After releasing the virtio-queue
crate with the relaxed version you can go back to using crates.io
One solution would be to make the
virtio-vsock
crate use thevirtio-queue
crate in the repo instead. This is how thevirtio-blk
crate does it.My PR now has a patch that does this. After releasing the
virtio-queue
crate with the relaxed version you can go back to using crates.io
In this case, can you also update to >=0.8.0? 0.8.0 is fixing a vm-memory bug that can affect the vsock crate, and it would be better to enforce 0.8.0 here.
This allows the use of newer versions of vm-memory and matches with other crates (e.g. vhost, linux-loader, etc) who also have such a relaxed vm-memory dependency.
Signed-off-by: Rob Bradford robert.bradford@intel.com