diff --git a/datasette/app.py b/datasette/app.py
index e7a5ce48..79692ff3 100644
--- a/datasette/app.py
+++ b/datasette/app.py
@@ -68,9 +68,7 @@ class BaseView(RenderMixin):
         self.ds = datasette
         self.files = datasette.files
         self.jinja_env = datasette.jinja_env
-        self.executor = datasette.executor
         self.page_size = datasette.page_size
-        self.max_returned_rows = datasette.max_returned_rows
 
     def options(self, request, *args, **kwargs):
         r = response.text('ok')
@@ -91,7 +89,7 @@ class BaseView(RenderMixin):
 
     async def pks_for_table(self, name, table):
         rows = [
-            row for row in await self.execute(
+            row for row in await self.ds.execute(
                 name,
                 'PRAGMA table_info("{}")'.format(table)
             )
@@ -135,49 +133,6 @@ class BaseView(RenderMixin):
             return name, expected, should_redirect
         return name, expected, None
 
-    async def execute(self, db_name, sql, params=None, truncate=False, custom_time_limit=None):
-        """Executes sql against db_name in a thread"""
-        def sql_operation_in_thread():
-            conn = getattr(connections, db_name, None)
-            if not conn:
-                info = self.ds.inspect()[db_name]
-                conn = sqlite3.connect(
-                    'file:{}?immutable=1'.format(info['file']),
-                    uri=True,
-                    check_same_thread=False,
-                )
-                self.ds.prepare_connection(conn)
-                setattr(connections, db_name, conn)
-
-            time_limit_ms = self.ds.sql_time_limit_ms
-            if custom_time_limit and custom_time_limit < self.ds.sql_time_limit_ms:
-                time_limit_ms = custom_time_limit
-
-            with sqlite_timelimit(conn, time_limit_ms):
-                try:
-                    cursor = conn.cursor()
-                    cursor.execute(sql, params or {})
-                    if self.max_returned_rows and truncate:
-                        rows = cursor.fetchmany(self.max_returned_rows + 1)
-                        truncated = len(rows) > self.max_returned_rows
-                        rows = rows[:self.max_returned_rows]
-                    else:
-                        rows = cursor.fetchall()
-                        truncated = False
-                except Exception:
-                    print('ERROR: conn={}, sql = {}, params = {}'.format(
-                        conn, repr(sql), params
-                    ))
-                    raise
-            if truncate:
-                return rows, truncated, cursor.description
-            else:
-                return rows
-
-        return await asyncio.get_event_loop().run_in_executor(
-            self.executor, sql_operation_in_thread
-        )
-
     def get_templates(self, database, table=None):
         assert NotImplemented
 
@@ -192,6 +147,7 @@ class BaseView(RenderMixin):
             as_json = kwargs.pop('as_json')
         except KeyError:
             as_json = False
+        table = kwargs.get('table', None)
         extra_template_data = {}
         start = time.time()
         status_code = 200
@@ -229,6 +185,23 @@ class BaseView(RenderMixin):
                         dict(zip(columns, row))
                         for row in rows
                     ]
+            elif '_shape' in request.args:
+                # Re-shape it
+                shape = request.raw_args['_shape']
+                if shape in ('objects', 'object'):
+                    columns = data.get('columns')
+                    rows = data.get('rows')
+                    if rows and columns:
+                        data['rows'] = [
+                            dict(zip(columns, row))
+                            for row in rows
+                        ]
+                        if shape == 'object' and table:
+                            pks = await self.pks_for_table(name, table)
+                            data['rows'] = {
+                                path_from_row_pks(row, pks, not pks): row
+                                for row in data['rows']
+                            }
             headers = {}
             if self.ds.cors:
                 headers['Access-Control-Allow-Origin'] = '*'
@@ -292,7 +265,7 @@ class BaseView(RenderMixin):
         extra_args = {}
         if params.get('_sql_time_limit_ms'):
             extra_args['custom_time_limit'] = int(params['_sql_time_limit_ms'])
-        rows, truncated, description = await self.execute(
+        rows, truncated, description = await self.ds.execute(
             name, sql, params, truncate=True, **extra_args
         )
         columns = [r[0] for r in description]
@@ -326,7 +299,6 @@ class IndexView(RenderMixin):
         self.ds = datasette
         self.files = datasette.files
         self.jinja_env = datasette.jinja_env
-        self.executor = datasette.executor
 
     async def get(self, request, as_json):
         databases = []
@@ -441,7 +413,7 @@ class RowTableShared(BaseView):
                     placeholders=', '.join(['?'] * len(ids_to_lookup)),
                 )
                 try:
-                    results = await self.execute(database, sql, list(set(ids_to_lookup)))
+                    results = await self.ds.execute(database, sql, list(set(ids_to_lookup)))
                 except sqlite3.OperationalError:
                     # Probably hit the timelimit
                     pass
@@ -504,17 +476,17 @@ class TableView(RowTableShared):
         if canned_query is not None:
             return await self.custom_sql(request, name, hash, canned_query['sql'], editable=False, canned_query=table)
         pks = await self.pks_for_table(name, table)
-        is_view = bool(list(await self.execute(name, "SELECT count(*) from sqlite_master WHERE type = 'view' and name=:n", {
+        is_view = bool(list(await self.ds.execute(name, "SELECT count(*) from sqlite_master WHERE type = 'view' and name=:n", {
             'n': table,
         }))[0][0])
         view_definition = None
         table_definition = None
         if is_view:
-            view_definition = list(await self.execute(name, 'select sql from sqlite_master where name = :n and type="view"', {
+            view_definition = list(await self.ds.execute(name, 'select sql from sqlite_master where name = :n and type="view"', {
                 'n': table,
             }))[0][0]
         else:
-            table_definition = list(await self.execute(name, 'select sql from sqlite_master where name = :n and type="table"', {
+            table_definition = list(await self.ds.execute(name, 'select sql from sqlite_master where name = :n and type="table"', {
                 'n': table,
             }))[0][0]
         use_rowid = not pks and not is_view
@@ -562,7 +534,7 @@ class TableView(RowTableShared):
         # _search support:
         fts_table = None
         fts_sql = detect_fts_sql(table)
-        fts_rows = list(await self.execute(name, fts_sql))
+        fts_rows = list(await self.ds.execute(name, fts_sql))
         if fts_rows:
             fts_table = fts_rows[0][0]
 
@@ -638,7 +610,7 @@ class TableView(RowTableShared):
         if request.raw_args.get('_sql_time_limit_ms'):
             extra_args['custom_time_limit'] = int(request.raw_args['_sql_time_limit_ms'])
 
-        rows, truncated, description = await self.execute(
+        rows, truncated, description = await self.ds.execute(
             name, sql, params, truncate=True, **extra_args
         )
 
@@ -678,7 +650,7 @@ class TableView(RowTableShared):
             # Attempt a full count, if we can do it in < X ms
             if count_sql:
                 try:
-                    count_rows = list(await self.execute(name, count_sql, params))
+                    count_rows = list(await self.ds.execute(name, count_sql, params))
                     filtered_table_rows = count_rows[0][0]
                 except sqlite3.OperationalError:
                     # Almost certainly hit the timeout
@@ -689,7 +661,7 @@ class TableView(RowTableShared):
 
         async def extra_template():
             display_columns, display_rows = await self.display_columns_and_rows(
-                name, table, description, rows, link_column=not is_view, expand_foreign_keys=True
+                name, table, description, rows, link_column=not is_view, expand_foreign_keys=not is_view
             )
             return {
                 'database_hash': hash,
@@ -755,8 +727,8 @@ class RowView(RowTableShared):
         params = {}
         for i, pk_value in enumerate(pk_values):
             params['p{}'.format(i)] = pk_value
-        # rows, truncated, description = await self.execute(name, sql, params, truncate=True)
-        rows, truncated, description = await self.execute(name, sql, params, truncate=True)
+        # rows, truncated, description = await self.ds.execute(name, sql, params, truncate=True)
+        rows, truncated, description = await self.ds.execute(name, sql, params, truncate=True)
         columns = [r[0] for r in description]
         rows = list(rows)
         if not rows:
@@ -813,7 +785,7 @@ class RowView(RowTableShared):
             for fk in foreign_keys
         ])
         try:
-            rows = list(await self.execute(name, sql, {'id': pk_values[0]}))
+            rows = list(await self.ds.execute(name, sql, {'id': pk_values[0]}))
         except sqlite3.OperationalError:
             # Almost certainly hit the timeout
             return []
@@ -826,7 +798,8 @@ class RowView(RowTableShared):
         foreign_key_tables = []
         for fk in foreign_keys:
             count = foreign_table_counts.get((fk['other_table'], fk['other_column'])) or 0
-            foreign_key_tables.append({**fk, **{'count': count}})
+            if count:
+                foreign_key_tables.append({**fk, **{'count': count}})
         return foreign_key_tables
 
 
@@ -835,8 +808,8 @@ class Datasette:
             self, files, num_threads=3, cache_headers=True, page_size=100,
             max_returned_rows=1000, sql_time_limit_ms=1000, cors=False,
             inspect_data=None, metadata=None, sqlite_extensions=None,
-            template_dir=None, static_mounts=None):
-        self.files = files
+            template_dir=None, static_mounts=None, memory=None):
+        self.files = list(files)
         self.num_threads = num_threads
         self.executor = futures.ThreadPoolExecutor(
             max_workers=num_threads
@@ -852,6 +825,90 @@ class Datasette:
         self.sqlite_extensions = sqlite_extensions or []
         self.template_dir = template_dir
         self.static_mounts = static_mounts or []
+        self.memory = memory or []
+        for filepath in self.memory:
+            if filepath not in self.files:
+                self.files.append(filepath)
+
+    async def initialize_memory_connections(self):
+        print('initialize_memory_connections')
+        for name, info in self.inspect().items():
+            print(name, info['memory'])
+            if info['memory']:
+                await self.execute(name, 'select 1')
+
+    async def execute(self, db_name, sql, params=None, truncate=False, custom_time_limit=None):
+        """Executes sql against db_name in a thread"""
+        def sql_operation_in_thread():
+            conn = getattr(connections, db_name, None)
+            if not conn:
+                info = self.inspect()[db_name]
+                if info['file'] in self.memory:
+                    # Copy file into an in-memory database
+                    memory_conn = sqlite3.connect(
+                        'file:{}?mode=memory&cache=shared'.format(db_name),
+                        uri=True,
+                        check_same_thread=False,
+                    )
+                    self.prepare_connection(memory_conn)
+                    # Do we need to copy data across?
+                    if not memory_conn.execute("select count(*) from sqlite_master where type='table'").fetchone()[0]:
+                        conn = sqlite3.connect(
+                            'file:{}?immutable=1'.format(info['file']),
+                            uri=True,
+                            check_same_thread=False,
+                        )
+                        self.prepare_connection(conn)
+                        pages_todo = conn.execute('PRAGMA page_count').fetchone()[0]
+                        print('Here we go...')
+                        i = 0
+                        for line in conn.iterdump():
+                            memory_conn.execute(line)
+                            i += 1
+                            if i % 10000 == 0:
+                                pages_done = memory_conn.execute('PRAGMA page_count').fetchone()[0]
+                                print('Done {}/{} {:.2f}%'.format(
+                                    pages_done, pages_todo, (pages_done / pages_todo * 100)
+                                ))
+                        conn.close()
+                    conn = memory_conn
+                else:
+                    conn = sqlite3.connect(
+                        'file:{}?immutable=1'.format(info['file']),
+                        uri=True,
+                        check_same_thread=False,
+                    )
+                    self.prepare_connection(conn)
+                setattr(connections, db_name, conn)
+
+            time_limit_ms = self.sql_time_limit_ms
+            if custom_time_limit and custom_time_limit < self.sql_time_limit_ms:
+                time_limit_ms = custom_time_limit
+
+            with sqlite_timelimit(conn, time_limit_ms):
+                try:
+                    cursor = conn.cursor()
+                    cursor.execute(sql, params or {})
+                    if self.max_returned_rows and truncate:
+                        rows = cursor.fetchmany(self.max_returned_rows + 1)
+                        truncated = len(rows) > self.max_returned_rows
+                        rows = rows[:self.max_returned_rows]
+                    else:
+                        rows = cursor.fetchall()
+                        truncated = False
+                except Exception:
+                    print('ERROR: conn={}, sql = {}, params = {}'.format(
+                        conn, repr(sql), params
+                    ))
+                    raise
+            if truncate:
+                return rows, truncated, cursor.description
+            else:
+                return rows
+
+        return await asyncio.get_event_loop().run_in_executor(
+            self.executor, sql_operation_in_thread
+        )
 
     def app_css_hash(self):
         if not hasattr(self, '_app_css_hash'):
@@ -975,6 +1032,10 @@ class Datasette:
                     'views': views,
 
                 }
+        # Annotate in-memory databases
+        memory_names = {Path(filename).stem for filename in self.memory}
+        for name, info in self._inspect.items():
+            info['memory'] = name in memory_names
         return self._inspect
 
     def app(self):
@@ -1020,4 +1081,6 @@ class Datasette:
             RowView.as_view(self),
             '/<db_name:[^/]+>/<table:[^/]+?>/<pk_path:[^/]+?><as_json:(\.jsono?)?$>'
         )
+        if self.memory:
+            app.add_task(self.initialize_memory_connections)
         return app
diff --git a/datasette/cli.py b/datasette/cli.py
index 5744d00a..055d6f92 100644
--- a/datasette/cli.py
+++ b/datasette/cli.py
@@ -233,7 +233,8 @@ def package(files, tag, metadata, extra_options, branch, template_dir, static, *
 @click.option('-m', '--metadata', type=click.File(mode='r'), help='Path to JSON file containing license/source metadata')
 @click.option('--template-dir', type=click.Path(exists=True, file_okay=False, dir_okay=True), help='Path to directory containing custom templates')
 @click.option('--static', type=StaticMount(), help='mountpoint:path-to-directory for serving static files', multiple=True)
-def serve(files, host, port, debug, reload, cors, page_size, max_returned_rows, sql_time_limit_ms, sqlite_extensions, inspect_file, metadata, template_dir, static):
+@click.option('--memory', type=click.Path(exists=True), help='database files to load into memory', multiple=True)
+def serve(files, host, port, debug, reload, cors, page_size, max_returned_rows, sql_time_limit_ms, sqlite_extensions, inspect_file, metadata, template_dir, static, memory):
     """Serve up specified SQLite database files with a web UI"""
     if reload:
         import hupper
@@ -262,6 +263,7 @@ def serve(files, host, port, debug, reload, cors, page_size, max_returned_rows,
         sqlite_extensions=sqlite_extensions,
         template_dir=template_dir,
         static_mounts=static,
+        memory=memory,
     )
     # Force initial hashing/table counting
     ds.inspect()