Repository navigation
server: an Add's image is the row its INSERT answers with - #10
Merged
Merged
Conversation
A recorder that asks for an Add's image is handed the row as the database stored it -- its precision, its defaults, what a trigger made of it -- and that was read back with a query of its own after the INSERT. One recorder that asks for every write's image is a trail of them, and then every Add was two statements: lesomnus/payday's trail is that recorder, and Rove's import pays it on every row it writes. The INSERT reads it now, with ent's `sql/returning` (protobuf-orm/ent#9): the generated Add asks the create builder for the row it stored, and where the dialect has RETURNING that is the INSERT's own answer. Where it has not, ent reads it with a query, which is what this did on every engine. The Add still answers with the row as the builder had it. The edges are not loaded, and need not be. A list edge of a row that did not exist a statement ago has nothing in it, and every other edge's key is in the row, which `Proto()` answers it from. `ent generate` needs `sql/returning` beside `sql/modifier` now; the README says so, and internal/apptest generates with both. - An Add whose recorder asks for its image sends no SELECT, on SQLite and PostgreSQL; it sent one. - The image of an Add is at the database's precision, which a clock with nanoseconds shows on PostgreSQL: unchanged, and now read by the INSERT. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01KxtnpRW533bdErnW8QPhsk
The pin was #9's head while it was open; it is master now, which has #9 and #8 beside it. Nothing it generates moved. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01KxtnpRW533bdErnW8QPhsk
lesomnus
added a commit
to lesomnus/payday
that referenced
this pull request
Oct 17, 2026
The two were pinned to the heads of protobuf-orm/ent#9 and protobuf-orm/protoc-gen-orm-ent#10 while those were open; both are merged, and these are the merge commits. Nothing generated moved but the ent runtime's record of its own version. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01KxtnpRW533bdErnW8QPhsk
lesomnus
added a commit
to lesomnus/rove
that referenced
this pull request
Oct 17, 2026
…tements Three more of lesomnus/payday#84, with what they need from protobuf-orm (protobuf-orm/ent#8 and #9, protobuf-orm/protoc-gen-orm-ent#10), and roster 67781c3 (lesomnus/roster#131), generated on the same payday: - payday#87: a line of the trail holds the row its write handed over. An Add's INSERT answers with it (`sql/returning`), and a Patch or an Apply hands over the row it read back to answer with; the trail read every write's row again. - payday#88: a domain transaction's trail is written five hundred rows to a statement and by its commit, where each row was a statement of its own -- and ahead of any statement about the trail in the same transaction, so an erasure that blanks a person's rows in the transaction that wrote them blanks every one of them. - payday#89: what the gate and the sink were told is the transaction's own, so `Deps.run` no longer begins with `seen.For`, and the line is gone. An imported row sent 36.9 statements and sends 18.8 (#103). `TestAnImportedRowIsAFewStatements` holds it at 22 a row, where it held 40. internal/ent's create builders have `Stored`, since `pd gen` generates with `sql/returning`; server/bare's Add asks for it, and server/pd is regenerated. No schema moved. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01KxtnpRW533bdErnW8QPhsk
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
1 participant
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.
A recorder that asks for an Add's image is handed the row as the database stored it: its precision, its defaults, and what a trigger made of it. That row was read back with a query of its own after the INSERT. lesomnus/payday's trail is a recorder that holds every write's row, so with it every Add is two statements. Rove's import pays that on every row it writes (lesomnus/payday#84).
What changes
Stored, ent'ssql/returning, sqlgraph, entc/gen: a create reads back the row it stored, by its INSERT where it can ent#9) when a recorder wants the image.RETURNING(PostgreSQL, SQLite), the INSERT answers with the row itself: one statement instead of two.Proto()answers it from.ent generateneedssql/returningbesidesql/modifier. The README says so, andinternal/apptestgenerates with both.internal/apptestpins the fork at its master, c860506, which has #9 merged.What it saves
In Rove, together with payday's trail asking for the row (lesomnus/payday,
recorder-takes-the-row): 36.9 statements per imported row, down to 27.8.Tests
testSnapshotCosts: an Add whose recorder asks for its image sends no SELECT on SQLite or PostgreSQL. It sent one before this change.TestPostgresRecorderSnapshotsSerializeWritersis unchanged and passes: an Add's image is at the database's precision, which a clock with nanoseconds shows on PostgreSQL. The INSERT now reads that image.buf generatetwice,go generateininternal/apptest(nothing moved but this change), then the root,runtimeandinternal/apptestsuites, the last on PostgreSQL 17 as well.🤖 Generated with Claude Code
https://claude.ai/code/session_01KxtnpRW533bdErnW8QPhsk