Use supportedFeatures object for feature detection

[wilzbach]

The new JSON output for -Xi=compilerInfo was added in #7838 and is still not documented and thus we can use the opportunity to make breaking changes.

Imho, it's a better idea to group all features in one object as there might be more coming up in the future.

CC @marler8997 @timotheecour

[dlang-bot]

Thanks for your pull request, @wilzbach!

Bugzilla references

Your PR doesn't reference any Bugzilla issue.

If your PR contains non-trivial changes, please reference a Bugzilla issue or create a manual changelog.

[jacob-carlborg]

The new JSON output for -Xi=compilerInfo was added in #7838 and is still not documented and thus we can use the opportunity to make breaking changes.

Did that PR end up in 2.079.0, because the PR contains a changelog entry.

[marler8997]

Yes

[wilzbach]

make[1]: *** [../generated/osx/release/32/dmd-unittest] Segmentation fault: 11 (core dumped)

Oh great the random OSX 32-bit segfaults strike again :/

[jacob-carlborg]

Is there a core dump available?

[wilzbach]

Is there a core dump available?

Nope and no ssh either. Despite the coredump wouldn't help much as it's known that 2.068.2 has bugs.
I think @marler8997 or @timotheecour already had a lot of fun trying to workaround them?

[jacob-carlborg]

Nope and no ssh either. Despite the coredump wouldn't help much as it's known that 2.068.2 has bugs.

Those wouldn’t be fixable in a 2.068.3 release?

[marler8997]

I was going to create a PR for 2.068 but the plan was to update the autotester version...

[wilzbach]

Yes that has been agreed upon here: https://forum.dlang.org/post/zbxaviowooiaenaawmgs@forum.dlang.org

AFAICT the agreement was the latest DMD compiler + LDC + gdc. The latter two are enforced by SemaphoreCI, so we could update auto-tester ... or well @braddr could.

[braddr]

I see no agreement in that thread. I see no statement of direction. I see a lot of differing opinions and contradictory statements of what can and should be done. No one from dlang leadership has stepped up and said that and what changes need to be made.

[JinShil]

No one from dlang leadership has stepped up and said that and what changes need to be made.

@WalterBright @andralex Does "dlang leadership" mean you? If so, can we get a decision about upgrading the autotester?

[andralex]

This is about the build infra, @MartinNowak would be the decider.

[wilzbach]

I see no agreement in that thread. I see no statement of direction.

From Martin's post in the post I linked:

We've been through bootstrapping discussions a couple of times, so let me repeat what was decided when we made the frontend switch from C++ to D.

• Other platforms are bootstrapped by cross-compilation.
• DMD must be compilable with the latest stable/major release of dmd, ldc, and gdc.

To enforce this policy the Travis-CI test was set up.
Hopefully this original purpose of the Travis-CI hasn't been forgotten in the meantime.
No other guarantees were negotiated.

(the travis tests have now been moved to Semaphore)

So while gdc is still at 2.068, I don't see anything blocking you from upgrading the auto-tester to newer versions. SemaphoreCI will ensure that DMD is still bootstrap-able with the latest release of the other compilers are still compilable

(and who knows maybe there's a GDC release soon - CC @ibuclaw, but from how I understand it that's independent from upograding the auto-tester hosts.)

[marler8997]

Hmmm, looks like this commit is the most likely culprit of the OSx failures...

[ibuclaw]

So while gdc is still at 2.068, I don't see anything blocking you from upgrading the auto-tester to newer versions. SemaphoreCI will ensure that DMD is still bootstrap-able with the latest release of the other compilers are still compilable

So essentially you can use whatever version you want, gdc will continue to ensure that there's a feature freeze on the compiler implementation.