-
Notifications
You must be signed in to change notification settings - Fork 1k
panic: runtime error: index out of range #649
Comments
Definite bad bookkeeping there. Shouldn't be possible to be visiting a dep that isn't depended on by anything. |
Its reproducible on CI: https://circleci.com/gh/weaveworks/common/133 But not on my mac... |
if that's a difference, then the first thing that jumps to mind is import path case sensitivity. i'm not specifically aware of a condition that could result this, but i'm generally aware that this is a problem area for us. any chance you've got some inconsistent casing in your import statements? |
A quick hack tells me no:
|
yeah probably not - i mean, this COULD be a thing:
b/c he has changed his username to lowercase, though that really should be all transparent and totally not be relevant here. without actually looking at the code, i'm stumped 😦 |
If I go for a lowecase sirupsen, I get a different error:
https://circleci.com/gh/weaveworks/common/136 And I can do an ensure locally anymore:
But I guess thats because some of my vendored dependancies depend on the uppercase version, so both are getting pulled in. |
yeah, that'll definitely do it. i had a long chat with someone in slack about this a while ago, i think we actually came to a reasonable, not-terrible solution. didn't make an issue for it, though 😢
ok, we have to have had a regression, these stale cache errors are suddenly popping up again all over the place. |
It does indeed seem to be something wrong with golang.org/x/sys. With verbose ensuring:
|
arrrrrggghhhh. ok, well, that at least gives us a ballpark. thanks 😄 |
Okay, if I try and use an older version of golang.org/x/sys, I get:
|
And its something about this PR, as master is still ensuring fine: https://circleci.com/gh/weaveworks/common/140 |
This looks to be the same / related to #639 |
I've got it to not fail by reverting all my changes to the lock file and just updated grpc: weaveworks/common#37 Hopefully thats helpful. |
Yes, i'm hoping #639 is the same problem. Thanks for all the info - it probably will be helpful, though i can't tackle this right now. Hopefully one of our intrepid contributors can. This may be difficult to replicate, given that it has to do with stale caches on disk, but we pretty clearly have an issue where local git repos aren't having the latest versions fetched down. #513 was the last place we fixed this, but I guess we've had a regression somehow. |
Seen just now in CI, haven't investigated yet:
https://circleci.com/gh/weaveworks/common/132
The text was updated successfully, but these errors were encountered: