Bug #3205
openGo compiler net test failing
Description
A recent commit appears to have broken the net test for the Go compiler:
https://build.golang.org/log/58be31cfd1a92ba9582fdf33e01f79e03184e59b
This was working on commit be02f354 and started failing when I upgraded to b7d3e1.
Updated by tuxillo about 5 years ago
- Status changed from New to In Progress
- Assignee set to tuxillo
Can you provide the steps to reproduce it? I'll try to bisect it. Also, are you sure you've updated world+kernel to latest?
Updated by Anonymous about 5 years ago
I'm not on HEAD, but I'm only 5 commits behind and they look unrelated. I'll try that though. I'll get you the steps to reproduce.
Updated by Anonymous about 5 years ago
Steps to reproduce:
1. Install go compiler from dports
2. git clone https://go.googlesource.com/go
3. cd go/src
4. ./make.bash
5. ../bin/go test -count 1 net
Updated by Anonymous about 5 years ago
When I upgraded to b7d3e1, I did a full world+kernel build/install and ran pkg upgrade -f after (no problems).
Updated by dillon about 5 years ago
It looks like GO is using a generator for the structural elements in the route messages. The elements have changed.
./vendor/golang.org/x/net/route/message.go
./vendor/golang.org/x/net/route/zsys_dragonfly.go
You may be able to fix it temporarily based on the comments at the top of the dragonfly file. We are looking into how to fix the base package permanently. The route structure did change in master but it shouldn't have in release.
-Matt
Updated by tuxillo about 5 years ago
Okay, I'll handle it and let you guys know. Hopefully we won't overlap ourselves.
Updated by tuxillo about 5 years ago
- Status changed from In Progress to Feedback
- % Done changed from 0 to 90
They pushed our changes to x/net, see:
https://github.com/golang/net/commit/c5a3c61f89f3ed696ec36b629ef1b97541165225
We've seen other failures too in 'net', unrelated to this. Were all those already present?
Updated by Anonymous about 5 years ago
The crypto/tls test reported in http://bugs.dragonflybsd.org/issues/3183 is still failing randomly and it's still not clear whether this is a bug in the test or in DF or both.