feat: Adds flag to optionally use system certificates when fetching latest release.

[mchernicoff]

Resolves #132 See that issue for details.

This adds the "native" flag to the axoupdater CLI. If this flag is set, axoupdater will fetch the latest release using native system certs for TLS, instead of the reqwest crate's default Mozilla WebPKI certs.

[mchernicoff]

This patch has not been fully tested in a production-like environment. In order to test it, a new release of axoupdater must first be created with these changes. Then you should run cargo dist build with cargo-dist with the updater option enabled to install axoupdater, then check that the newly created update command correctly allows the use of the "native" flag.

[mistydemeo]

Apologies for the delay on reviewing here.

This makes sense to me conceptually, but I'm wondering a bit about the usability. In what circumstances would a user know that they need to pass the --native flag? Is this something that users in your environment are going to pass explicitly, or a default that all users using your distributed software will need? If it's the latter, I wonder if it would be better to have the CLI gate this based on environment variables.

[mchernicoff]

Apologies for the delay on reviewing here.

This makes sense to me conceptually, but I'm wondering a bit about the usability. In what circumstances would a user know that they need to pass the --native flag? Is this something that users in your environment are going to pass explicitly, or a default that all users using your distributed software will need? If it's the latter, I wonder if it would be better to have the CLI gate this based on environment variables.

To answer the question as asked (even as other solutions are being implemented), we don't expect most users of our software to have to use this flag. The issue is that we, the primary developers and maintainers of the software, need this feature to make our own software work in our corporate enterprise environment. So I was envisioning something that we, the developers, know to use and that other users can be told to use if they encounter the same error we get (e.g. A line in the README somewhere).

[mistydemeo]

I believe this was resolved by #136, which was released in 0.6.7 - does that fully cover your usecase?

[mchernicoff]

I believe this was resolved by #136, which was released in 0.6.7 - does that fully cover your usecase?

Yes, we should be good now. Thank you!