[dependabot]

Bumps github.com/oschwald/maxminddb-golang/v2 from 2.5.0 to 2.8.0.

Release notes

Sourced from github.com/oschwald/maxminddb-golang/v2's releases.

2.8.0

• String caching now uses buckets to improve concurrent decoding. On 64-bit systems, the table grows from 64 KiB to 72 KiB and is allocated only when an eligible string is decoded. Entry capacity grows from 4,096 to 7,168, which can also increase retained string memory. Pull request by James Ranson. GitHub #252.
• Standalone mmdbdata.NewDecoder calls now default to no string cache. This reduces setup cost for short-lived decoders. Use the new mmdbdata.WithStringCache() option for repeated reads through a retained decoder or its cursors. Without the option, repeated string reads can allocate more and run slower. Cursors supplied by a Reader continue to share its caching policy.
• Added mmdbdata.Decoder.CursorAt(offset) to read independent records from one buffer without changing the decoder's position. The cursors share its string cache and can be read concurrently while the buffer remains valid and unchanged.
• Reduced maxminddb-gen generation and test time by reusing loaded dependencies during output validation, avoiding duplicate syntax work, and omitting unused parser metadata. Full type checking and generated-output validation remain enabled. Validation normalizes paths from package loaders to avoid false duplicate declarations. GitHub #256.
• Reduced reflection decoding overhead for string struct fields, including fields read through database pointers, while retaining existing validation and custom unmarshaling behavior.
• Shared string-cache handling between map keys and decoded string values.
• Fixed Reader.Verify() accepting invalid UTF-8 values overwritten by duplicate map keys. Verification now checks every encoded string in the data and metadata sections, including overwritten values.
• Fixed maxminddb-gen rejecting fields that use type aliases of supported types, including aliases in nested containers.
• Fixed maxminddb-gen panicking when a byte slice or its elements use an imported type alias.
• Fixed maxminddb-gen emitting inaccessible type names for imported container aliases with private element or map key types. Generated helpers infer those types while retaining allocation, overflow, and size checks.
• Fixed maxminddb-gen emitting private byte aliases in constraints for nested byte slice aliases.
• Fixed maxminddb-gen emitting inaccessible types for imported aliases whose generic arguments contain private types, fields, or methods.

2.7.0

• Go 1.26 or later is now required. CI now tests Go 1.26 and 1.27.
• Fixed Reader.Verify() rejecting a search-tree record that points to a value nested in another data record. The MaxMind DB spec permits this, and mmdbwriter can write such databases. GitHub #250.
• Decoding now rejects extended type bytes 0 and 250 through 255, which the MaxMind DB spec does not define. Before, they decoded as other types. For example, 0x00 0xfb decoded as a string.
• Reader.Verify() now rejects a data pointer that points into the middle of a field.

... (truncated)

Changelog

Sourced from github.com/oschwald/maxminddb-golang/v2's changelog.

2.8.0 - 2026-10-11

• String caching now uses buckets to improve concurrent decoding. On 64-bit systems, the table grows from 64 KiB to 72 KiB and is allocated only when an eligible string is decoded. Entry capacity grows from 4,096 to 7,168, which can also increase retained string memory. Pull request by James Ranson. GitHub #252.
• Standalone mmdbdata.NewDecoder calls now default to no string cache. This reduces setup cost for short-lived decoders. Use the new mmdbdata.WithStringCache() option for repeated reads through a retained decoder or its cursors. Without the option, repeated string reads can allocate more and run slower. Cursors supplied by a Reader continue to share its caching policy.
• Added mmdbdata.Decoder.CursorAt(offset) to read independent records from one buffer without changing the decoder's position. The cursors share its string cache and can be read concurrently while the buffer remains valid and unchanged.
• Reduced maxminddb-gen generation and test time by reusing loaded dependencies during output validation, avoiding duplicate syntax work, and omitting unused parser metadata. Full type checking and generated-output validation remain enabled. Validation normalizes paths from package loaders to avoid false duplicate declarations. GitHub #256.
• Reduced reflection decoding overhead for string struct fields, including fields read through database pointers, while retaining existing validation and custom unmarshaling behavior.
• Shared string-cache handling between map keys and decoded string values.
• Fixed Reader.Verify() accepting invalid UTF-8 values overwritten by duplicate map keys. Verification now checks every encoded string in the data and metadata sections, including overwritten values.
• Fixed maxminddb-gen rejecting fields that use type aliases of supported types, including aliases in nested containers.
• Fixed maxminddb-gen panicking when a byte slice or its elements use an imported type alias.
• Fixed maxminddb-gen emitting inaccessible type names for imported container aliases with private element or map key types. Generated helpers infer those types while retaining allocation, overflow, and size checks.
• Fixed maxminddb-gen emitting private byte aliases in constraints for nested byte slice aliases.
• Fixed maxminddb-gen emitting inaccessible types for imported aliases whose generic arguments contain private types, fields, or methods.

2.7.0 - 2026-09-29

• Go 1.26 or later is now required. CI now tests Go 1.26 and 1.27.
• Fixed Reader.Verify() rejecting a search-tree record that points to a value nested in another data record. The MaxMind DB spec permits this, and mmdbwriter can write such databases. GitHub #250.
• Decoding now rejects extended type bytes 0 and 250 through 255, which the MaxMind DB spec does not define. Before, they decoded as other types. For example, 0x00 0xfb decoded as a string.

... (truncated)

Commits

• a495bc9 Set release version and date
• aa0e63e Merge pull request #260 from oschwald/perf/generator-type-info
• 2e0ae68 Normalize file paths during generated-package validation
• d77548f Merge main into generator performance branch
• 1c70308 Reduce unused parser work in maxminddb-gen
• 9ce6d1e Reuse loaded packages when validating generated decoders
• df4444b Merge pull request #258 from oschwald/dependabot/go_modules/golang.org/x/tool...
• f1b833d Merge pull request #259 from oschwald/perf/string-field-decoding
• 0e60381 Support type aliases in generated decoder fields
• 7cc1919 Validate overwritten strings during database verification
• Additional commits viewable in compare view

[zakird]

@dependabot recreate