proposal: x/sync/errgroup: propagate panics through Go

[qiulaidongfeng]

Proposal Details

See #53757 (comment) , #53757 can delay the propagation of panics arbitrarily far into the future, making mistakes that used to be promptly discovered latent.

We can panic during Group.add to preserve the original behavior(fast panic when with not call Wait).
However, this means that we are in Group.Go propagate panic.
This is API change ,so I submit this proposal.

I can send CL implement this proposal.

open-ended question：
Is propagate runtime.Goexit through Go?
Propagate the first panic or all panics?
A subtle difference from the previous fast panic, where the previous panic had to be recovered within the function, now it can be recovered outside of the function. Is this acceptable?

[qiulaidongfeng]

To clarify, https://go.dev/cl/682335 It can alleviate this problem #53757 (comment) ,but not solve it.

[gabyhelp]

Related Issues

• x/sync/errgroup: propagate panics and Goexits through Wait #53757 (closed)
• proposal: x/sync: add panicgroup struct to deal with panic in the sub-goroutine #49802 (closed)

_{(Emoji vote if this was helpful or unhelpful; more detailed feedback welcome in this discussion.)}

[gopherbot]

Change https://go.dev/cl/682935 mentions this issue: errgroup: revert propagation of panics

[seankhliao]

Since we're reverting #53757 this is going to be a won't do.