ISqlDialect gains MaxBindParameters — the per-command ceiling the seam was
missing while modelling everything else that differs per backend: 2098 usable
of SQL Server's 2100 (sp_executesql spends two), 1000 for Oracle's IN-list
limit (ORA-01795, tighter than its 64k bind count), 65535 for the Postgres
Bind message.
SqlGroupPlanner.AddKeyValuePlans now emits several plans for one group when
its distinct keys exceed that ceiling, instead of one statement the provider
refuses after the connection opened — which left every tag in an EAV group
publishing BadCommunicationError on every poll, permanently, and reported the
database endpoint as down. Chunks share the group key; members follow their
own key so a shared keyValue never splits. SqlPollReader is unchanged: it runs
one command per plan and slices each plan's own Members back, proven by a
fixture read at a ceiling of 1.
NAMEDATALEN-1 (63) and Oracle 12.2's 128 are byte limits in the database
encoding, not character counts, so a 32-character Cyrillic name passed
ident.Length and was then silently truncated (Postgres) or rejected by the
server (Oracle). Both dialects now measure Encoding.UTF8.GetByteCount and say
'bytes' in the rejection. SqlServerDialect deliberately keeps ident.Length —
sysname is nvarchar(128), a real character limit — pinned by a control test.
Found by the new OracleReadTests against a real Oracle 23 Free: EVERY Oracle
driver instance failed its Initialize-time catalog load, and every AdminUI browse
of a table's columns threw, with
ORA-01745: invalid host/bind variable name
Oracle rejects a bind variable named after a reserved word, and the catalog
readers bind a parameter whose base name is `table` (SqlCatalogLoader /
SqlBrowseSession bind `schema` + `table`), so `ListColumnsSql` could not execute
at all. The provider was structurally non-functional.
OracleDialect now namespaces every marker/bind name through one `p_` constant
(`:p_k0` / `p_k0`), which a reserved word can never collide with, and its catalog
SQL is respelled to match. Prefixing everything rather than carrying a
reserved-word list is deliberate: the list is long and version-dependent, and a
future base name (`date`, `user`, `comment`, `level`…) would silently re-open the
same hole. Marker and bind name are derived from the same constant, so they
cannot drift apart.
No offline test could have caught this — only a server parses the SQL — so the
regression test asserts the invariant that made it possible (a bind name is never
the bare base name), with `table` called out explicitly.
Live evidence: 6/6 OracleReadTests red with ORA-01745 before, 6/6 green after,
against gvenzl/oracle-free:23-slim.
Joseph Doherty