refactor: elimate string manipulation in TacticAnalysis.verifyTryThisSuggestions

[chenson2018]

This change should be compatible with, but independent from, #37916. Instead of manipulating pretty-printed strings to identify when grind? uses hashes or approx, search for the corresponding syntax kinds.

---

[github-actions]

PR summary 9b51fe6ec0

Import changes for modified files

No significant changes to the import graph

Import changes for all files

Files	Import difference

---

Declarations diff

+ approx

You can run this locally as follows

## summary with just the declaration names:
./scripts/pr_summary/declarations_diff.sh <optional_commit>

## more verbose report:
./scripts/pr_summary/declarations_diff.sh long <optional_commit>

The doc-module for scripts/pr_summary/declarations_diff.sh contains some details about this script.

---

No changes to technical debt.

You can run this locally as

./scripts/reporting/technical-debt-metrics.sh pr_summary

• The relative value is the weighted sum of the differences with weight given by the inverse of the current value of the statistic.
• The absolute value is the relative value divided by the total sum of the inverses of the current values (i.e. the weighted average of the differences).

[joneugster]

I couldn't find any tests about verifyTryThisSuggestions in MathlibTest/TacticAnalysis.lean. Could you add some so I can test this easier?

Concretely I do see tests about set_option linter.tacticAnalysis.verifyGrind true in but I didn't see one which would trigger the modified "[...] suggestion failed: [...]" warning.

[joneugster]

I don't yet understand why removing the pretty-printing is desired? But maybe an added test makes this obvious

[chenson2018]

I feel like this is inherently hacky, I believe added by Claude when first written. It's just using the pretty-printed tactic to check for hash codes and approx in grind =>. Besides being less maintainable and clear in my opinion, I think it will skip any usage of grind that uses a declaration with the substring "approx".

[chenson2018]

I couldn't find any tests about verifyTryThisSuggestions in MathlibTest/TacticAnalysis.lean. Could you add some so I can test this easier?

Concretely I do see tests about set_option linter.tacticAnalysis.verifyGrind true in but I didn't see one which would trigger the modified "[...] suggestion failed: [...]" warning.

I have added a test of a suggestion failed warning that would be incorrectly skipped given the current use of strings versus the typed syntax. I'm not sure that any testing around this will be great however, as it's checking behavior that in the hopefully near future will be resolved in core.

[chenson2018]

-awaiting-author