Motivation: rust-nostr 0.45 introduced a broad local-relay hardening series whose effective defaults were mostly absent from its changelog and entirely absent from ngit-grasp's custom NIP-11 response. One new 64 KiB event bound is incompatible with production history containing a valid roughly 149 KiB NIP-34 patch event. Leaving other defaults implicit risks another dependency upgrade silently changing serving policy, while exposing every internal knob would create configuration that peers cannot usefully negotiate. Approach: explicitly select every retained rust-nostr hardening value in the builder. Expose only the subscription and per-filter result limits that peers can discover and ngit-grasp sync already consumes, plus the Git-specific event size policy. Apply one filter-limit option consistently to explicit, query, and omitted-limit caps. Raise the event default from 64 KiB to 192 KiB and validate it remains beneath the fixed 5 MiB WebSocket message ceiling. Publish the standard NIP-11 limitation fields and correct the architecture/reference documentation, including removal of the obsolete max_filters claim. Correctness: the NIP-11 max_subscriptions value feeds the existing per-session subscription ledger; default_limit feeds adaptive pagination with its existing verification-page safeguard. max_limit is advertised truthfully but is not misused as an omitted-filter promise. The 192 KiB bound clears the observed patch by about 29% while preserving a finite allocation boundary. All three new options are synchronized across source, reference docs, NixOS, and the environment example. Excluded scope: message-size negotiation, filter-payload limits, per-IP fairness, and non-standard NIP-11 extensions remain separate work. Rate, handshake, subscription-memory, filter-count, and negentropy bounds are pinned but deliberately not operator-configurable because our sync cannot negotiate them through standard NIP-11 fields. Validation: focused unit tests passed for explicit configuration defaults, event/WebSocket size validation, configured NIP-11 advertisement, and the full http::nip11::tests module (10 tests before adding the focused override case). The previously validated full library suite and deployment build were not repeated at the user's request.
2.6 KiB
Embedded relay limits
ngit-grasp embeds rust-nostr LocalRelay 0.45.0. The application selects every
effective limit explicitly so future dependency defaults cannot silently alter
production admission policy.
Effective limits
| Limit | ngit-grasp default | Operator configuration | NIP-11 |
|---|---|---|---|
| Total inbound connections | Unbounded | NGIT_MAX_CONNECTIONS |
No standard field |
| Active REQs per connection | 500 | NGIT_RELAY_MAX_SUBSCRIPTIONS |
max_subscriptions |
| Results per filter | 500 | NGIT_RELAY_FILTER_LIMIT |
max_limit, default_limit |
| Serialized event size | 192 KiB | NGIT_RELAY_MAX_EVENT_SIZE_BYTES |
No equivalent field |
| Event writes per minute | 60 | Fixed | No standard field |
| Queries per minute | 120 | Fixed | No standard field |
| Authentication events per minute | 30 | Fixed | No standard field |
| Text messages per minute | 300 | Fixed | No standard field |
| WebSocket message size | 5 MiB | Fixed | max_message_length |
| Handshake deadline | 10 seconds | Fixed | No standard field |
| Subscription-ID length | 250 bytes | Fixed | max_subid_length |
| Filters per REQ | 20 | Fixed | No standard field |
| Subscription state per connection | 1 MiB | Fixed | No standard field |
| Active negentropy sessions per connection | 10 | Fixed | No standard field |
| Negentropy items per connection | 50,000 | Fixed | No standard field |
| Negentropy frame | 60,000 bytes | Fixed upstream | No standard field |
The filter setting is applied consistently to rust-nostr's explicit filter cap, per-query result cap, and omitted-limit default. Limits are per filter; results from multiple filters in one REQ are merged without an aggregate truncation.
The 192 KiB event default is three times rust-nostr's new 64 KiB default. Production history contains a valid NIP-34 patch event of about 149 KiB, so 64 KiB is incompatible with ngit-grasp's purpose. The raised limit remains bounded and below the 5 MiB WebSocket message ceiling.
Client adaptation
ngit-grasp refetches NIP-11 per connection session. Its outbound sync ledger
uses max_subscriptions, falling back conservatively when absent. Adaptive
historic pagination uses default_limit with a verification page and learns
from raw delivered page sizes. max_limit describes explicit filter limits;
historic sync currently omits limit, so it does not treat max_limit as an
omitted-filter page-size promise.
NIP-11 has no standard max_filters field in the current schema. Filter-count
and serialized-message budgets therefore remain conservative client-side
constants rather than falsely negotiated capabilities.