Skip to content

chore(deps): update dependency @tailor-platform/sdk to v2.40.0 - #228

Merged
k1LoW merged 1 commit into
mainfrom
renovate/tailor-platform-sdk-and-erp-kit-packages
Nov 13, 2026
Merged

k1LoW merged 1 commit into
mainfrom
renovate/tailor-platform-sdk-and-erp-kit-packages

Conversation

@renovate

@renovate renovate Bot commented Nov 12, 2026 •

Copy link
Copy Markdown
Contributor

This PR contains the following updates:

Package Change Age Confidence
@tailor-platform/sdk (source) 2.38.0 → 2.40.0 age confidence

Release Notes

tailor-platform/sdk (@​tailor-platform/sdk)

v2.40.0

Compare Source

Minor Changes
  • #​2986 57fa11a Thanks @​dqn! - files in db, resolver, executor, workflow, and httpAdapter now treats a pattern starting with ! as an exclusion: for each file, the last files pattern that matches it decides. Before, such a pattern matched nothing and excluded nothing, so a config that already lists ! patterns now leaves those files out, and the next deploy removes the resources they defined. A file that several files patterns match is now loaded once.

    ignores is deprecated and will be removed in v3, together with the default exclusion of **/*.test.ts and **/*.spec.ts. tailor upgrade runs the v3/files-ignores-to-negation codemod, which moves ignores into files as ! patterns and adds "!**/*.test.ts" and "!**/*.spec.ts" to services without ignores.

    export default defineConfig({
      name: "my-app",
      executor: {
        files: ["executors/**/*.ts", "!executors/ci-only.ts"],
      },
    });
  • #​3025 ee30e41 Thanks @​dqn! - The plugin hook onNamespaceLoaded is renamed to onTailorDBNamespaceLoaded, which says that it runs once for each TailorDB namespace in the config (except external ones). Its context and output types are renamed to TailorDBNamespaceLoadedContext and TailorDBNamespacePluginOutput.

    onNamespaceLoaded, PluginNamespaceProcessContext, and NamespacePluginOutput are deprecated and will be removed in v3; plugins that use them keep working until then. A plugin that defines both onNamespaceLoaded and onTailorDBNamespaceLoaded is rejected when the config loads. tailor upgrade runs the v3/tailordb-namespace-loaded-rename codemod, which renames the hook and the two types.

    export const auditLogPlugin: Plugin = {
      id: "@example/audit-log",
      description: "Adds an audit log table",
      importPath: "@example/audit-log",
      onTailorDBNamespaceLoaded({ namespace }) {
        return { tables: { auditLog: db.table(`AuditLog_${namespace}`, { message: db.string() }) } };
      },
    };
  • #​3014 3af9d37 Thanks @​dqn! - A resolver's permission can now compare an input with the caller through an { input: "<field>" } operand, e.g. [{ input: "employeeId" }, "=", { user: "id" }], so a caller cannot act on another user's data by sending their ID. The operand names a top-level non-array t.string(), t.uuid(), or t.enum() input field, is compared with =, can only be used in a permit: true policy, and never matches when the input is omitted. defaultPermission cannot use it.

  • #​3028 3093df4 Thanks @​dqn! - Add --watch to tailor function run: it runs the function again each time you save the function file, a project file it imports, or the argument file, without deploying, and prints each run as --> / <-- lines tagged with the execution ID. Add --arg-file to read the argument from a JSON file; without it, --watch creates a temporary argument file and prints its path. --arg is now validated before anything contacts the Platform, so invalid JSON fails even for a resolver without input.

  • #​2943 fd6655b Thanks @​toiroakr! - With TypeScript 7.1 or later, add the SDK as a content mapper for .tailor files in tsconfig.json and TypeScript reads Env, Attributes, MachineUserName, IdpName, and the other names your tailor.config.ts defines directly from the config, so you no longer need tailor.d.ts. When the tsconfig.json that applies to tailor.config.ts enables the mapper, tailor generate and tailor deploy stop writing tailor.d.ts and warn if an old copy remains.

  • #​3017 06e8d17 Thanks @​dqn! - Export the TailorAnyField type so helpers that build t.object() inputs or outputs can accept any t.* or db.* field and keep its type.

  • #​2991 5bc1ffd Thanks @​dqn! - Breaking change (beta): mockSecretmanager now makes getSecret() and getSecrets() reject with a vault not found: <vault> error for a vault it was not given, as they do on the Platform, instead of resolving undefined or {}. A missing secret in an existing vault still resolves to undefined and is omitted from getSecrets(). A test that reads a vault it never seeded now fails; seed the vault with mockSecretmanager({ secrets: { "<vault>": { ... } } }), or with {} to test a missing secret. tailorRuntime({ config }) seeds only the vaults declared in defineSecretManager(), so also seed vaults created with tailor secret vault create and the vaults tailor deploy creates for IdP clients. setSecrets() replaces and reset() clears config-loaded vaults too, so a test that calls them and then reads a config vault now fails; use setSecret() or mergeSecrets() to add values while keeping those vaults.

Patch Changes
  • #​3027 20b75aa Thanks @​dqn! - tailor deploy now deletes the application of a config that no longer defines or references any TailorDB namespace, resolver namespace, IdP, or Auth service. Before, the application kept the services it was last deployed with, so it kept serving them, and a deploy that dropped one of those services deleted its tables or resolvers and then failed with ... is used by applications. A config already in that state deletes its leftover application on the next deploy. deploy asks for confirmation before deleting it, so a non-interactive run needs --yes.

  • #​2994 60bc80b Thanks @​dqn! - TailorDB migration diffs, and the schema drift details tailor function run reports, now show foreign key changes on existing fields, such as ~ ownerId: foreignKey: User.id → Person.id, instead of omitting them or printing configuration changed. In the schema drift details, a default value or create hook added to or removed from a field now shows as optionalOnCreate: false → true or optionalOnCreate: true → false.

  • #​3011 5fcc3ce Thanks @​dqn! - Fix the "Event Payloads" section of the executor guide: it names the exported *Args trigger args types instead of interfaces the SDK does not export, no longer lists a scheduledTime field that ScheduleArgs does not have, and its resolver-executed examples narrow on success so they type-check.

  • #​3012 ade7982 Thanks @​dqn! - Report an invalid Secret Manager vault or secret name, HTTP adapter name, AI Gateway authNamespace, or tailor workflow executions --workflow-name value with the same message as OAuth2 and IdP client names: "must be 3-63 lowercase letters, digits, or hyphens, starting and ending with a letter or digit". Secret Manager names previously reported only the regular expression they failed to match.

  • #​3024 7a8bd33 Thanks @​dqn! - Stop saying that removed fields lose their data in the tailor tailordb migration resolve --rollback warning and in the suggestions from tailor deploy and tailor tailordb migration validate (namespace without a migration checkpoint) and tailor tailordb migration rebaseline (workspace tables that differ from the latest migration). They now say the fields are removed and only the deleted tables lose their data. The TailorDB migration guide also notes that a field removed by tailor tailordb migration sync may come back with its earlier values when a migration adds it again.

  • #​2993 9634c2b Thanks @​dqn! - Internal refactoring: remove an unused field from how tailor deploy plans workflow deletions. No user-facing behavior change.

  • #​2971 fbaf9b7 Thanks @​toiroakr! - tailor deploy now stops with SERVICE_IN_USE before it applies anything when the deploy would delete an Auth service that an application still references as external. Before, such a deploy deleted the Auth service's machine users and then failed on the Auth service itself, because the Platform refuses to delete an Auth service another application uses, which left the workspace half applied. This covers applications in the same deploy and applications that the deploy does not include. Keep the Auth service in the config that defines it, or move the referencing application to another Auth service in the same deploy.

  • #​3023 1a201fa Thanks @​dqn! - tailor remove now also stops with SERVICE_IN_USE before it deletes anything when another application still references an Auth service it would delete, and both tailor deploy and tailor remove stop the same way for a TailorDB namespace or resolver namespace that another application references as external. The error lists every application that uses each service. Before, they deleted the namespace's tables (with their data) or resolvers, or the Auth service's machine users, and then failed on the namespace or Auth service itself, because the Platform refuses to delete one that another application uses, which left the workspace half applied. Keep the resource in the config that defines it, or first remove the referencing applications or deploy them without it.

  • #​3009 52dafad Thanks @​renovate! - fix(deps): update dependency temporal-polyfill to v1.0.8

  • #​3022 1684419 Thanks @​renovate! - chore(deps): update dependency graphql to v17.1.0

  • #​2976 c24496d Thanks @​dqn! - Add copilot, claude, and codex to tailor setup ci autofix --codemod. Each works as deterministic, then hands what the codemods leave for review, with their migration prompts, to that AI coding agent in a job of its own, tailor-codemod-agent, which installs and runs none of the project's code. The fix commit carries the agent's edits as chore: apply the SDK upgrade with <agent>, leaving out .github/workflows and, with claude, the configuration files claude-code-action resets; if the agent fails, it carries only the codemods' rewrites. The agent runs on each push to a pull request from the repository while changes for review remain, except right after the workflow's own fix commit and for bots other than renovate[bot]; the warnings keep listing every change that was left for review. Claude and Codex read an API key from the TAILOR_AUTOFIX_ANTHROPIC_API_KEY or TAILOR_AUTOFIX_OPENAI_API_KEY repository secret. Copilot reads a fine-grained personal access token from TAILOR_AUTOFIX_COPILOT_TOKEN, or, with --copilot-auth github-token, signs in with the workflow token. The optional TAILOR_AUTOFIX_CLAUDE_MODEL, TAILOR_AUTOFIX_CODEX_MODEL, or TAILOR_AUTOFIX_COPILOT_MODEL repository variable picks the agent's model. tailor setup ci env lists the secret, if any, and the variable, and tailor setup update keeps both flags. With autofix-ci and securefix, an agent value moves the service's commit into the tailor-autofix-push job. When tailor upgrade and tailor generate still change files right after the workflow's own fix commit, the error says the agent's edits may be the cause, and on a pull request from the repository a tailor-autofix-comment job, with pull-requests: write, posts it as a pull request comment.

  • #​2989 b6fedd2 Thanks @​dqn! - Fix tailor deploy, tailor tailordb migration resolve, and tailor tailordb migration sync not explaining how to proceed when records with no value for a field block changing it from optional to required.

  • #​3026 fc25c9f Thanks @​dqn! - TailorDB migration diffs, and the schema drift details tailor function run reports, now show the same change details for a modified nested member as for a top-level field, such as ~ address: members: ~zip (required: false → true), instead of only its path.

v2.39.0

Compare Source

Minor Changes
  • #​2958 a955502 Thanks @​toiroakr! - defineAuth results now have asExternal(), which references an Auth service defined in another app of the same repository as an external one: auth: coreAuth.asExternal(). It takes over the name and the machine users, so the referencing app follows renames of both and gets the machine user names in its generated tailor.d.ts. A misspelled invoker there is then a type error, which { name: "...", external: true } could not provide because it leaves the names empty. Hooks, OAuth2 clients, and the rest of the definition are not taken over, so the referencing app does not deploy them again.

  • #​2952 93c373f Thanks @​toiroakr! - tailor deploy now stops with INVOKER_MACHINE_USER_NOT_FOUND when a string invoker of a resolver or an executor names a machine user that does not exist in its Auth service. Before, a misspelled name passed tsc and deploy, and failed with machine user not found when the resolver or executor first ran. When a config deployed in the same run defines the Auth service, the machine users it lists are the ones accepted, because deploy removes the ones it no longer lists. When the Auth service is only referenced as external, the machine users that exist in the workspace are accepted. An existing config that names a machine user that exists nowhere now fails at deploy; add the machine user to the Auth service or fix the name. If you create the machine user of an externally referenced Auth service after deploying the app that uses it, deploy the Auth service first, or deploy both configs in the same run (tailor deploy --config a,b).

  • #​2975 5889f05 Thanks @​jackchuka! - Breaking change: tailor workspace prune no longer accepts --personal. Every workspace now belongs to an organization, so there are no workspaces outside an organization and folder for it to select. Replace --personal with --organization-root or --folder-id naming the locations to sweep. Do not just delete it: a --name or --older-than sweep without a location considers every workspace visible to the login, even with --yes.

  • #​2739 8b69e57 Thanks @​toiroakr! - tailor seed apply now runs the TailorDB seed and the IdP _User seed, list and delete scripts as temporary workflows instead of synchronous script executions, so they are no longer cut off by the 60-second execution limit. For each namespace it registers a function, a job function and a workflow named tailor-seed-<kind>--<namespace> (<kind> is tailordb, idp-seed, idp-list or idp-truncate), runs every chunk through them, and removes them afterwards. A namespace too long for the platform's 63-character name limit is cut short and followed by a hash. A run interrupted before cleanup leaves them behind, and the next run of the same script (the TailorDB seed, or the IdP _User seed, list or delete) for that namespace removes them first.

    If tailor seed apply finds the workflow of an earlier run still executing (for example, a chunk that keeps running on the platform after Ctrl-C), it stops with an error naming the execution instead of removing it. tailor seed apply --truncate runs this check for every seed workflow of the namespaces it touches before it deletes anything, so an interrupted run no longer leaves a truncated table that is not reseeded. Wait for that execution to finish and run the command again. Running tailor seed apply for the same namespace in two processes at once is not supported: if one of them fails while registering the function, it removes the resources of the other. If the command loses track of a running chunk (for example, a timeout while starting it), it removes the resources and cannot tell whether the chunk was applied, so check the table, then re-run with --upsert.

    A TailorDB chunk that fails now stops the remaining chunks of that namespace, so later chunks do not fail on foreign keys of the rolled-back chunk or leave a gap in a table split across chunks.

    The seed and seed dump scripts are now minified and, like other bundled functions, embed an inline sourcemap by default so error stack traces stay readable. Set buildOptions.inlineSourcemap to false in defineConfig to drop the sourcemap.

Patch Changes
  • #​2981 1a4dca2 Thanks @​dqn! - tailor executor trigger --data now delivers the JSON to an incoming-webhook executor as its body, as a request to the webhook URL does. Since a Platform change, the data had arrived one level deeper, under body.body, so reading a field such as body.email returned undefined without an error. If an executor reads body.body to work around this, change it back to body. --header values are not passed to the executor, which receives empty headers, and the command now warns when --header is given. --header without --data no longer sends a body, so the executor receives body as null, as with no options. To test an executor that reads headers, send the request to its webhook URL, which tailor executor webhook list shows. The programmatic triggerExecutor passes payload to the executor as body unchanged, so pass the body itself rather than { body, headers }.

  • #​2959 ebb62b6 Thanks @​dqn! - tailor generate and tailor function run now read the config path from TAILOR_CONFIG_PATH when --config is not given, like other commands that take a single config file. If TAILOR_CONFIG_PATH holds comma-separated paths for tailor deploy, pass --config to these commands.

  • #​2960 61dbc34 Thanks @​dqn! - tailorRuntime({ config }) now names the file, line, and parse error when tailor.config.ts or a file it imports does not parse, instead of reporting [object Object] against tailor.config.ts.

  • #​2983 d4faf97 Thanks @​dqn! - Document that an executor with an incoming webhook trigger, when triggered manually with tailor executor trigger or triggerExecutor, receives no rawBody and empty headers, in the executor guide and in the JSDoc of the webhook rawBody and headers args. An executor that verifies a signature over rawBody rejects such a run, so test it through its webhook URL.

  • #​2957 b79b113 Thanks @​dqn! - Fix tailor deploy --no-schema-check leaving a table removed from the local definitions without a migration until the next deploy when its namespace has pending migrations. The table is now deleted in the same deploy, after those migrations complete.

  • #​2941 7a5c244 Thanks @​dqn! - tailor deploy now prints the pending-migration summary and the other notices about the migration state (a checkpoint reset, tables left to delete, a migration to run again, a skipped script, ignored steps) once, before the plan, instead of repeating them after the confirmation. When the pending migrations change after the plan, for example because another deploy applied one, the deploy now stops with DEPLOY_PLAN_STALE instead of applying a different set than the one it showed.

  • #​2987 67f8539 Thanks @​dqn! - tailor query --engine sql and query() from @tailor-platform/sdk/cli now report how many rows an INSERT, UPDATE, or DELETE changed. For these statements the result adds affectedRows and sets rowCount to the same count, which was 0 when the statement had no RETURNING. When such a statement returns no rows, the text output prints the count (for example "3 rows affected.") instead of "No rows returned."

  • #​2988 818debc Thanks @​dqn! - Speed up tailor seed apply and tailor tailordb migration test for a seed table too large for one chunk: splitting it took tens of seconds for a few MB of small records and now takes milliseconds

  • #​2961 f30a654 Thanks @​dqn! - Add --codemod to tailor setup ci autofix, so the autofix workflow handles the codemods of a pull request that upgrades @tailor-platform/sdk, such as one from Renovate. It upgrades each app from the version the base branch's lockfile resolves, or from 2.37.0 when the base branch has no such lockfile, to the version the pull request installs. warn reports what tailor upgrade would rewrite and the changes it leaves for review, error reports them as errors and fails the tailor-codemod-result job, which also fails when the autofix job fails or is cancelled, and deterministic commits the rewrites with what tailor generate changes and reports the rest. tailor setup update keeps the --codemod value, and without --codemod the workflow is the same as before. With error, a review that still matches after you finish the migration keeps the check failing until the base branch has the new version. deterministic leaves files under .github/workflows out of the fix commit and reports them for you to commit, and a bun repository needs bun.lock instead of bun.lockb.

  • #​2947 d372640 Thanks @​renovate! - fix(deps): update dependency kysely to v0.29.8

  • #​2951 ef7e152 Thanks @​renovate! - fix(deps): update @​​connectrpc to v2.3.0

  • #​2954 d065035 Thanks @​renovate! - fix(deps): update dependency @​​opentelemetry/semantic-conventions to v1.44.0

  • #​2955 0e86776 Thanks @​renovate! - fix(deps): update dependency @​​politty/zod to v0.7.1

  • #​2956 5f2bfc5 Thanks @​renovate! - fix(deps): update dependency globals to v17.14.0

  • #​2973 c0ae913 Thanks @​renovate! - fix(deps): update dependency semver to v7.8.6

  • #​2978 4dc187c Thanks @​renovate! - fix(deps): update oxc to v0.157.0

  • #​2997 d396c09 Thanks @​renovate! - chore(deps): lock file maintenance

  • #​2739 8b69e57 Thanks @​toiroakr! - tailor seed apply now runs each seed chunk in a single transaction. If any table fails, every table inserted earlier in that chunk is rolled back instead of being left committed, and Executors triggered by create events no longer run mid-chunk, which could deadlock on self-referencing tables.

    Seed data is split into chunks of about 3.5 MiB, and the transaction covers one chunk. A table that does not fit in one chunk is split across chunks and is committed chunk by chunk.

  • #​2984 eb76277 Thanks @​dqn! - Labeling a closed pull request, such as one already merged, no longer deploys a preview workspace from the preview workflow that tailor setup ci preview --require label generates. The cleanup workflow had already run when the pull request closed, so such a workspace stayed until a later cleanup run deleted it after its 7-day ttl expired. The tailor-preview-deploy job of every preview workflow now skips events on a closed pull request, and your jobs that need it are skipped with it.

    This release also raises the workflow template version, so tailor setup check reports a template-version finding for existing workflows until you run tailor setup update. With TAILOR_PLATFORM_FAIL_ON_DRIFT set to true, the drift check fails until then.

  • #​2982 cd33cdb Thanks @​dqn! - The v2 migration guide and tailor upgrade prompt for the programmatic CLI functions no longer say that triggerExecutor's payload is typed from the executor definition. payload is accepted only for an executor with an incoming webhook trigger, and its shape is not checked; startWorkflow's arg is still typed from the workflow definition.


Configuration

📅 Schedule: (UTC)

  • Branch creation
    • At any time (no schedule defined)
  • Automerge
    • At any time (no schedule defined)

🚦 Automerge: Disabled by config. Please merge this manually once you are satisfied.

♻ Rebasing: Whenever PR becomes conflicted, or you tick the rebase/retry checkbox.

🔕 Ignore: Close this PR and you won't be reminded about this update again.


  • If you want to rebase/retry this PR, check this box

This PR was generated by Mend Renovate. View the repository job log.

@renovate
renovate Bot requested a review from a team as a code owner November 12, 2026 05:22
@renovate
renovate Bot force-pushed the renovate/tailor-platform-sdk-and-erp-kit-packages branch from 63aec3f to ebcbc1d Compare November 13, 2026 05:55
@renovate renovate Bot changed the title chore(deps): update dependency @tailor-platform/sdk to v2.39.0 chore(deps): update dependency @tailor-platform/sdk to v2.40.0 Nov 13, 2026
@k1LoW
k1LoW merged commit 01037a0 into main Nov 13, 2026
7 checks passed
@k1LoW
k1LoW deleted the renovate/tailor-platform-sdk-and-erp-kit-packages branch November 13, 2026 05:56
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant