From 0.13 to 0.14

0.14 changes the signature of every block method that works on a record. Update the calls below, then run better-supabase sql sync and generate a migration for the new comments and attachments author functions.

Block methods take a subject and read the tenant from the blockCopy Anchor Link

The methods of comments, attachments, tags, reactions, relations, versions, sla, share links, custom fields, reminders, time tracking, approvals and the state machine take the record they work on as a SubjectRef, { type, id }, as their first argument. They no longer take an organizationId: the blocks read it from their new tenant option, an organization id or a function that returns the current one. A call without a tenant fails with invalid_request and the hint TENANT_REQUIRED. The state machine and the sla clock methods take no tenant, since the record decides it.

Rows carry subject in place of subjectType and subjectId. Batch methods take a list of subjects and return a record keyed type:id, which subjectKey(subject) from better-supabase/blocks builds. Event payloads, the SQL functions and their columns keep subjectType and subjectId.

Before	After
comments.create({ organizationId, subjectType, subjectId, body })	comments.create(subject, { body })
comments.list(organizationId, subjectType, subjectId, options?)	comments.list(subject, options?)
comments.counts(organizationId, subjectType, subjectIds), keyed by id	comments.counts(subjects), keyed type:id
comments.copy(organizationId, from, to)	comments.copy(from, to)
comments.history(organizationId, subject?, options?)	comments.history(subject?, options?)
attachments.upload({ organizationId, subjectType, subjectId, ... })	attachments.upload(subject | null, attachment)
attachments.put({ organizationId, subjectType, subjectId, ... }, file)	attachments.put(subject | null, attachment, file)
attachments.list(organizationId, subjectType?, subjectId?)	attachments.list(subject?)
versions.snapshot({ organizationId, subjectType, subjectId, ... })	versions.snapshot(subject, input?)
versions.list(organizationId, subjectType, subjectId, options?)	versions.list(subject, options?)
shareLinks.create(organizationId, subjectType, subjectId, options?)	shareLinks.create(subject, options?)
shareLinks.recordView(organizationId, subjectType, subjectId, viewerKind, viewer?)	shareLinks.recordView(subject, viewerKind, viewer?)
shareLinks.list(subjectType, subjectId), setState(subjectType, subjectId, starred)	shareLinks.list(subject), setState(subject, starred)
tags.create(organizationId, kind, name, ...), tags.list(organizationId, kind?)	tags.create(kind, name, ...), tags.list(kind?)
tags.tag(organizationId, subjectType, subjectId, tagId), and untag	tags.tag(subject, tagId), and untag
tags.forSubject(organizationId, subjectType, subjectId)	tags.forSubject(subject)
reactions.react(organizationId, subjectType, subjectId, emoji, options?)	reactions.react(subject, emoji, options?)
reactions.summary(organizationId, subjectType, subjectIds), keyed by id	reactions.summary(subjects), keyed type:id
reactions.customEmoji(organizationId), saveCustomEmoji(organizationId, input)	reactions.customEmoji(), saveCustomEmoji(input)
customFields.listDefinitions(organizationId, entity), saveDefinition(organizationId, entity, ...)	customFields.listDefinitions(entity), saveDefinition(entity, ...)
customFields.readValues(organizationId, subjectType, subjectId)	customFields.readValues(subject)
customFields.setValue(organizationId, subjectType, subjectId, definitionId, value)	customFields.setValue(subject, definitionId, value)
relations.link({ organizationId, source, target, kind, ... }), and unlink	relations.link(source, target, { kind, data?, fields? }), and unlink(source, target, { kind })
relations.of(organizationId, subject, kind?), edges(organizationId, kind)	relations.of(subject, kind?), edges(kind)
relations.ancestors(organizationId, kind, subject), and descendants	relations.ancestors(subject, kind), and descendants
relations.rollup({ organizationId, subject, ... })	relations.rollup(subject, { kind, of, agg, column? })
relations.convertedInto(organizationId, origin, intoType), lineage(organizationId, subject)	relations.convertedInto(origin, intoType), lineage(subject)
sla.savePolicy(organizationId, policy)	sla.savePolicy(policy)
approvals.request({ organizationId, subjectType, subjectId, action, ... })	approvals.request(subject, { action, reason?, payload? })
approvals.list(organizationId, options?)	approvals.list(options?), with an optional subject filter
approvals.isApproved(subjectType, subjectId, action)	approvals.isApproved(subject, action)
stateMachine.transition(subjectType, subjectId, to, reason?)	stateMachine.transition(subject, to, reason?), and the same for pause, resume, setFlag, state, history and stages
reminders.save(organizationId, subjectType, subjectId, offsetMinutes?, fields?)	reminders.save(subject, offsetMinutes?, fields?)
reminders.list(organizationId, subjectType, subjectId)	reminders.list(subject)
timeTracking.start(organizationId, fields?), stop(organizationId), running(organizationId)	timeTracking.start(fields?), stop(), running()
timeTracking.log, list, timesheet, submit and timesheets with organizationId first	the same calls without it; entries take subject instead of subjectType and subjectId
row.subjectType, row.subjectId	row.subject.type, row.subject.id
useComments({ organizationId, subjectType, subjectId, schema?, topic?, limit? })	useComments({ subject, topic?, limit? })
useReactions({ organizationId, subjectType, subjectIds, schema?, topic? })	useReactions({ subjects, topic? })
useTagInput({ organizationId, subjectType, subjectId, kind, schema? })	useTagInput({ subject, kind })

To upgrade:

1. Pass tenant to the blocks you create yourself (createComments({ transport, tenant: () => ctx.tenant })), or build them with createBlockClient or blockSet.client, which pass the request's tenant (see one block per request).
2. Replace each organizationId, subjectType, subjectId argument list with one { type, id } object, and read row.subject instead of row.subjectType and row.subjectId. Read batch results by subjectKey(subject).
3. Pass blocks: blockConfig to the browser createClient, and tenant when the tenant isn't the session's tenant claim, then drop schema and organizationId from the hooks (see block hooks).
4. Run better-supabase sql sync, then generate a migration, for the comments and attachments author functions.

One block client per request instead of a wrapper per blockCopy Anchor Link

A lib/blocks.ts that wrapped each block to pass the organization can go. createBlockClient(ctx.db, commentsBlock, { config: blockConfig }) and blockSet.client(ctx.db, "comments") build one block for a request: they call through ctx.db and act in ctx.db.$context.tenant.

Comments and attachments return their authorCopy Anchor Link

comments.list and attachments.list return author on each row, with the name and avatar URL from the profiles module, through the new security definer functions comment_authors(uuid[]) and attachment_authors(uuid[]). They return only authors of rows the caller may read. Without the profiles module, author.name and author.avatarUrl are undefined.