Open raznar opened 8 months ago
For an example on why this is problematic, picture the following situation:
A
is consumed by package B
A
, on a completely unrelated symbol to on consumed within package B
. This change is tagged and released under v1.1.0
, and as apart of CI, scip is re-generated and published against the main
branchB
is a low touch package, meaning it's not released very often, and scip is not re-indexed with the symbols new versionA
, it will use 1.1.0
, but package B
has yet to re-index, and is still using the version 1.0.0
. This will result in no symbols returning from the find references searchWhile increasing the frequency that Package B
runs its indexing would technically address this above problem. At scale, with hundreds of packages, this problem essentially boils down to not being able to trust the results of find references at all
Having some way of running this find references search, while ignoring the version segment in the generated symbols, would allow us to rely more on the data that find references, and other precise intelligence functional can provide
Worth noting that I just submitted a feature request that would resolve this problem a bit of a different way: https://github.com/sourcegraph/sourcegraph/issues/60276
Feature request description
Would like the ability to track symbols across version updates, which allows end users to see how a symbol's use evolves over time.
Is your feature request related to a problem? If so, please describe.
Customer reports that cross-repo find references breaks when a version of a symbol has been updated but older versions of that symbol still exist. Product gap link
Describe alternatives you've considered.
None
Additional context
More information in this Slack thread