search

mozilla / mentat

UNMAINTAINED A persistent, relational store inspired by Datomic and DataScript.
https://mozilla.github.io/mentat/
Apache License 2.0
1.65k stars 115 forks source link

[edn] Add EDN differ #220

Open ncalexan opened 7 years ago

ncalexan commented 7 years ago

Similar to #195, it gets really tiresome trying to visually diff to subtly different edn::Value instances. This ticket tracks implementing some structured diff algorithm. Even just pretty printing and doing a text diff on the two outputs would be better than the madness I'm doing to compare right now!

victorporof commented 7 years ago

I have interest in working on this as well, possibly after #195. If there's anyone else interested in taking this though, please feel free.

victorporof commented 7 years ago

This can either be done via structural diff, or textual diff. Structural is likely more difficult to get right, one can imagine a tree traversal iterator using itertools::diff_with, but that likely won't offer meaningful data when diffing two sufficiently different ASTs (need to bail out early). Textual diff seems like a reasonable approach, especially after #195

ncalexan commented 7 years ago

This is really an answer to https://github.com/mozilla/mentat/issues/247#issue-205518531, but I'm putting it here since this is the relevant ticket.

I want pretty output during tests -- basically, I want assert_eq! to help me find differences in EDN (presented as &str slices). This probably looks like a new macro, assert_eq_edn! or similar, that parses both sides into EDN (bonus points: handles edn::Value and anything that can be parsed into EDN) and then provides a helpful diff message as well as the string inputs if the two sides aren't equal.

victorporof commented 7 years ago

@ncalexan It would seem that textual diffing is something sufficient for your needs (and thus requiring some new pretty printing rules for json-like expanding of edn, which would also allow me to highlight the usefulness of some of the generalizations in #245). Is that correct?

ncalexan commented 7 years ago

On Thu, Feb 9, 2017 at 12:07 PM, Victor Porof notifications@github.com wrote:

@ncalexan https://github.com/ncalexan It would seem that textual diffing is something sufficient for your needs (and thus some new pretty printing rules for json-like expanding of edn, which would also allow me to highlight the usefulness of some of the generalizations in #245 https://github.com/mozilla/mentat/pull/245). Is that correct?

Textual diffing is great for test reports and helping humans figure out where things are broken. The EDN matching I describe is really about letting us write tests easily, which is slightly different. Progress on either front would be great!