1981747 - Add `./mach app-services setup`

Status: ASSIGNED

(Firefox Build System :: General, enhancement, P3)

Description

[Ben Dean-Kawamura [:bdk]]

Once https://bugzilla.mozilla.org/show_bug.cgi?id=1980203 is merged, then we should be able to run cargo test -p to test our app-services crates. However, for many of our crates, we still need to figure out how to link to NSS.

Let's create a mach command that sets up the app-services environment, similar to verify-desktop-environment.sh. It should:

• Run ./mach build to build NSS
• Set the NSS_DIR variable to point to the the built NSS in the object directory

I don't think it needs to do anything else at this point, but we may want to extend it to do more in the future.

Implementing this is an alternative to https://bugzilla.mozilla.org/show_bug.cgi?id=1958000, if we can make Cargo do the linking then we don't need to do anything special with mach rusttests.

Comment 1

[Ben Dean-Kawamura [:bdk]]

Attachment: Bug 1981747 - Add `./mach app-services setup`, r=glandium,markh

Techincally we could land this and it would work, but it doesn't really
make sense until app-services is in the moz-central tree.

Comment 2

[Ted Campbell [:tcampbell]]

Something to consider is using the third_party/python/gyp in tree for predictability, since at least firefox is building nss using that version.

Comment 3

[Ben Dean-Kawamura [:bdk]]

Would the idea to build NSS ourselves? I avoided that because I wanted to minimize the amount of code we needed to maintain, but maybe we should be building NSS. That would give us some more control.

Comment 4

[Ted Campbell [:tcampbell]]

oh sorry, I misread the initial context. I think using the desktop nss is reasonable, but my understanding is that it doesn't build the static nss parts (which is why I was looking at extending the build so it did).

Comment 5

[Ben Dean-Kawamura [:bdk]]

Yes that's correct and not using the static NSS does have some drawbacks. For example the pkix parts aren't present which makes the rc_crypto tests fail. I think this the current code is good enough for now, but I have a fast-follow issue for improving this: https://bugzilla.mozilla.org/show_bug.cgi?id=1989633