Repository navigation
perf: reuse GetTIDsByTokenExpr results across searches via fraction index cache - #573
Open
eguguchkin wants to merge 5 commits into
Open
eguguchkin wants to merge 5 commits into
eguguchkin wants to merge 5 commits into
Conversation
Codecov Report❌ Patch coverage is
Additional details and impacted files@@ Coverage Diff @@
## main #573 +/- ##
==========================================
+ Coverage 79.45% 79.47% +0.02%
==========================================
Files 253 255 +2
Lines 19152 19242 +90
==========================================
+ Hits 15217 15293 +76
- Misses 3935 3949 +14 ☔ View full report in Codecov by Harness. 🚀 New features to boost your workflow:
|
dkharms
reviewed
Oct 14, 2026
Comment on lines
+10
to
+14
| // ExprTIDs is a value of the ExprTIDs cache layer: TIDs matched by a | ||
| // scan-heavy token expression (wildcard, regex, range) of the immutable | ||
| // fraction. Key keeps the original parser.TokenKey string to detect | ||
| // uint32 hash collisions. Exact-match literals are resolved by binary | ||
| // search alone and never enter this cache. |
Comment on lines
+61
to
+64
| // ExprTIDs caches GetTIDsByTokenExpr results of the immutable | ||
| // fraction for scan-heavy token expressions (see ExprTIDs); | ||
| // keyed by the hash of parser.TokenKey, evicted by the | ||
| // shared cleaner. |
Comment on lines
+39
to
+42
| // exprTIDsCache is the ExprTIDs layer of the fraction's IndexCache; | ||
| // the fraction is immutable, so cached lookups are valid for the | ||
| // fraction's whole lifetime. | ||
| exprTIDsCache cache.Cache[ExprTIDs] |
Comment on lines
+297
to
+300
| func isExactLiteral(t parser.Token) bool { | ||
| lit, ok := t.(*parser.Literal) | ||
| return ok && len(lit.Terms) == 1 && lit.Terms[0].Kind == parser.TermText | ||
| } |
Member
There was a problem hiding this comment.
There is the same check in pattern.newLiteralSearch (you've mentioned it yourself):
func newLiteralSearch(base baseSearch, token *parser.Literal) *literalSearch {
if len(token.Terms) != 1 || token.Terms[0].Kind != parser.TermText {
return nil
}
/* ... */
}Let's make a separate method for parser.Literal.
Contributor
Author
There was a problem hiding this comment.
I just moved IsExactLiteral into the parser package but didn’t combine it with newLiteralSearch because their signatures are different.
Comment on lines
+294
to
+296
| // isExactLiteral reports whether t is a plain `field:value` literal — the | ||
| // only expression resolved without scanning tokens; see | ||
| // pattern.newLiteralSearch for the same condition. |
| // key, it is stable across independently parsed but equal queries. | ||
| // | ||
| // String parts are length-prefixed, so keys of different tokens cannot | ||
| // collide through concatenation. |
Member
There was a problem hiding this comment.
Again, too wordy comment.
It will be much better to have one sentence with examples:
// TokenKey returns a canonical string uniquely identifying
// the lookup semantics of the token:
// - First example;
// - Second example;
// - ...;
Contributor
There was a problem hiding this comment.
// TokenKey constructs an opaque canonical key describing everything a token
// lookup needs: the selected field, the token type, term kinds and data,
// range endpoints and inclusivity, canonical IP-range bounds, or the regex
// source. Equal keys denote interchangeable lookup behavior, including for
// leaf tokens produced by independently parsed but semantically equal queries.
//
// The key encodes string components with explicit lengths so adjacent fields
// and terms cannot collide by concatenation. Derived token state is omitted:
// parsing normalizes regex data, IP bounds, and term inputs first, ensuring
// the key depends only on their observable search semantics rather than AST
// identity or cached implementation details.
//
// Callers may compare, store, or hash the result, but must not interpret its
// internal layout or rely on key ordering.
Looks better to me
Contributor
Author
There was a problem hiding this comment.
i replaced implementation with DumpSeqQL
| // | ||
| // String parts are length-prefixed, so keys of different tokens cannot | ||
| // collide through concatenation. | ||
| func TokenKey(t Token) string { |
Member
There was a problem hiding this comment.
I wonder -- why did you decide to implement such function instead of adding DumpSeqQL into parser.Token interface or invoking DumpSeqQL directly on t?
var b strings.Builder
switch t := t.(type) {
case *Literal:
t.DumpSeqQL(&b)
case *Range:
t.DumpSeqQL(&b)
case *IPRange:
/* ... */
default:
panic(fmt.Sprintf("unknown token type: %T", t))
}Are you sure that thare are no problems with characters escaping in your approach?
dkharms
approved these changes
Oct 14, 2026
Contributor
🔴 Performance DegradationSome benchmarks have degraded compared to the previous run. Show table
|
This branch has not been deployed
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
4 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
If you have used LLM/AI assistance please provide model name and full prompt: