Normlize selector use within lotus

[ribasushi]

No description provided.

[codecov]

Codecov Report

Merging #7467 (a40aa9f) into master (1993efe) will decrease coverage by 0.03%.
The diff coverage is 91.66%.

@@            Coverage Diff             @@
##           master    #7467      +/-   ##
==========================================
- Coverage   39.21%   39.17%   -0.04%     
==========================================
  Files         631      631              
  Lines       66787    66788       +1     
==========================================
- Hits        26191    26165      -26     
- Misses      36024    36056      +32     
+ Partials     4572     4567       -5     

Impacted Files	Coverage Δ
node/impl/client/client.go	45.11% <91.66%> (+0.07%)	⬆️
journal/types.go	86.66% <0.00%> (-13.34%)	⬇️
chain/actors/builtin/miner/diff.go	48.52% <0.00%> (-10.30%)	⬇️
chain/events/observer.go	71.64% <0.00%> (-6.72%)	⬇️
extern/sector-storage/manager_calltracker.go	57.70% <0.00%> (-4.85%)	⬇️
chain/actors/builtin/paych/paych.go	14.77% <0.00%> (-4.55%)	⬇️
chain/exchange/peer_tracker.go	66.66% <0.00%> (-4.31%)	⬇️
node/hello/hello.go	63.63% <0.00%> (-3.41%)	⬇️
journal/fsjournal/fs.go	66.66% <0.00%> (-2.78%)	⬇️
itests/kit/blockminer.go	93.65% <0.00%> (-1.59%)	⬇️
... and 15 more

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 1993efe...a40aa9f. Read the comment docs.

[rvagg]

I've been slightly hesitant to do because it's not precisely the same selector, but functionally should be.

We could simplify further by cleaning up the textselector API I reckon, but that's not critical and I think we should work on integrating that functionality into go-ipld-prime and clarifying the API when doing that.

[ribasushi]

because it's not precisely the same selector

Wait... they should be identical:

• previous in lotus: https://github.com/filecoin-project/lotus/blob/v1.13.0-rc1/node/impl/client/client.go#L1304-L1306
• current in g-f-m: https://github.com/filecoin-project/go-fil-markets/blob/v1.13.1/shared/selectors.go#L13-L15
• what we replaced both with: https://github.com/ipld/go-ipld-prime/blob/v0.12.3/traversal/selector/parse/selector_parse.go#L88

What am I missing?

[rvagg]

I think it's just a simple matter of key ordering, I haven't spent more than a couple of brain cycles on parsing this, but ssb.ExploreRecursive(selector.RecursionLimitNone(), ssb.ExploreAll(ssb.ExploreRecursiveEdge())).Node() is:

{"R":{":>":{"a":{">":{"@":{}}}},"l":{"none":{}}}}

while CommonSelector_ExploreAllRecursively is:

{"R":{"l":{"none":{}},":>":{"a":{">":{"@":{}}}}}}

The difference alone was enough to make be baulk before diving in, but I think it's functionally the same, just comes out different when dag-jsonified.