would like `jq` or `json` available in the GZ

[davepacheco]

I think this will be important during support/service while debugging to help analyze logs and validate config files. I'm familiar with json but I think the world has probably chosen jq instead.

[jclulow]

Note also, for logs, that we're already shipping looker in the image today. For log files in particular (rather than configuration files) it's probably a better place to do filtering than jq because it, like bunyan before it, supports logs that include things that aren't strictly JSON (e.g., SMF logs). I would welcome additional issues filed for looker for things that are missing for support purposes there as well!

[davepacheco]

Hmm. The main thing I use json for is for pulling properties out of the JSON objects to do basic analysis. For example: oxidecomputer/omicron#3082 (comment)

I often pass the results to some combination of grep, awk, sort, or column. I'm all for adding json-like features to looker as well, but these tools (jq and json) do already do this and the JSON-only input is not a problem for this use because it's easy to filter the input for bunyan records (especially when you've already got a multi-stage pipeline with multiple greps).

(Of course, I think eventually we will want to build better higher-level tools for analyzing our logs. I expect it may be some time before we have that. Even then, we may find ourselves wanting lower-level tools in some contexts.)

[jclulow]

This should be sorted out now: we are shipping both looker and jq in the image.