Type mapping (T-SQL → PostgreSQL)

SQL Server	PostgreSQL	Notes
int / bigint / smallint	integer / bigint / smallint	direct
tinyint	smallint	PG has no tinyint
bit	boolean	defaults 0/1 → false/true
decimal(p,s) / money	numeric(p,s) / numeric(19,4)	direct
nvarchar(n) / nvarchar(max)	varchar(n) / text	byte halving for nvarchar
datetime / datetime2 / datetimeoffset	timestamp(3) / timestamp(6) / timestamptz	precision preserved
uniqueidentifier	uuid	requires pgcrypto
varbinary / image	bytea	direct
IDENTITY(1,1)	GENERATED BY DEFAULT AS IDENTITY	seed + step preserved; sequences reset to MAX(id) after copy

Default expression conversion

SQL Server default	PostgreSQL output
GETDATE() / SYSDATETIME()	now()
GETUTCDATE() / SYSUTCDATETIME()	(now() at time zone 'utc')
NEWID()	gen_random_uuid()
SUSER_SNAME()	current_user
DB_NAME() / HOST_NAME()	current_database() / inet_client_addr()::text
CONVERT([date], expr)	CAST(expr AS date)
'X'+expr on string columns	'X' || expr
DEFAULT 0 on datetime	TIMESTAMP '1900-01-01 00:00:00'

Smart constraint fallback

When a btree index hits the 2,704-byte row limit, the runner reacts:

Index shape	What happens
1 column, non-unique, no INCLUDE	Auto-retry as USING HASH ✓
Multi-column or has INCLUDE	Logged to {db}_constraints_report.sql — DBA decides (md5 expr, partial WHERE, GIN trigram)
UNIQUE oversized	Logged with a warning — hash would risk collisions
FK violated by source data (orphan rows)	Logged with the data-cleanup options inline
CHECK violated by source data	Logged; suggested NOT VALID rewrite

Unported objects inventory

Triggers, stored procedures, views, scalar/table-valued functions are not auto-rewritten (T-SQL → PL/pgSQL is too lossy). Instead:

• ✓ Captured into _migration_meta.objects on the target — queryable, filterable
• ✓ Original T-SQL preserved verbatim in tsql_definition
• ✓ Companion workbook file {db}_unconverted_objects.sql with PostgreSQL stubs above each block
• ✓ port_status column to track progress: PENDING / IN_PROGRESS / DONE / IGNORED
• ✓ View _migration_meta.unported shows what's left

CLI commands

a1mig inspect       Read source schema and print a summary report.
a1mig convert       Generate the PostgreSQL DDL script (writes to file).
a1mig schema        Phase 1: drop+create target DB and apply tables (no idx/FK).
a1mig data          Phase 2: copy rows (binary COPY, deferred FKs).
a1mig constraints   Phase 3: indexes + FKs + check constraints (smart fallback).
a1mig objects       Phase 4: capture triggers/procs/views/functions into
                    _migration_meta.objects + write porting workbook.
a1mig full          Run all phases in order, then verify.
a1mig verify        Compare row counts between source and target tables.