11/10/2022, 7:52 AM
I can understand it being a performance or stability issue if there was a recursive loop, but there are a bunch of cases where e.g. a framework or binary version is pointed at by a symlink to indicate 'latest'/'active'…

Stefano Bonicatti

11/10/2022, 2:28 PM
I don’t think we have a table that reads symlinks; I see we have the
table that detects if a file is a symlink or not. I think it might make sense to have an optional column where we use the
API (on Unix) to read the symlink target, but from there it’s a bit tricky. The symlink could point to another symlink; the
API always reads the “contents” of the first symlink but doesn’t expand it further. In that case if you want the path to be expanded until you find a file, we would have to put a depth limit. At the same time while the symlink may point to a real file, the path itself could be composed by symlinks; not sure if that’s something that you would need too.. but that’s likely even more tricky and potentially performance intensive depending on the number of symlinks to chase


11/10/2022, 3:09 PM
That's at least confirmation that there's no know workarounds! I may as well put the feature request in then, it's reasonable to expect only straightforward use cases to be achievable. Even one with limited capabilities would be helpful IMO