Re: review: FDW API

I wrote:
> ... My feeling is it'd be best to pass down
> all the information the executor node has got --- probably we should
> just pass the ForeignScanState node itself, and leave a void * in that
> for FDW-private data, and be done with it.  Otherwise we're going to be
> adding missed stuff back to the API every time somebody notices that
> their FDW can't do X because they don't have access to the necessary
> information.

Attached is a rewritten version of fdwhandler.sgml that specifies what I
think is a more future-proof API for the callback functions.  Barring
objections, I'll push ahead with editing the code to match.

            regards, tom lane


<!-- doc/src/sgml/fdwhandler.sgml -->

 <chapter id="fdwhandler">
   <title>Writing A Foreign Data Wrapper</title>

   <indexterm zone="fdwhandler">
    <primary>foreign data wrapper</primary>
    <secondary>handler for</secondary>
   </indexterm>

   <para>
    All operations on a foreign table are handled through its foreign data
    wrapper, which consists of a set of functions that the planner and
    executor call. The foreign data wrapper is responsible for fetching
    data from the remote data source and returning it to the
    <productname>PostgreSQL</productname> executor. This chapter outlines how
    to write a new foreign data wrapper.
   </para>

   <para>
    The FDW author needs to implement a handler function, and optionally
    a validator function. Both functions must be written in a compiled
    language such as C, using the version-1 interface.
    For details on C language calling conventions and dynamic loading,
    see <xref linkend="xfunc-c">.
   </para>

   <para>
    The handler function simply returns a struct of function pointers to
    callback functions that will be called by the planner and executor.
    Most of the effort in writing an FDW is in implementing these callback
    functions.
    The handler function must be registered with
    <productname>PostgreSQL</productname> as taking no arguments and returning
    the special pseudo-type <type>fdw_handler</type>.
    The callback functions are plain C functions and are not visible or
    callable at the SQL level.
   </para>

   <para>
    The validator function is responsible for validating options given in the
    <command>CREATE FOREIGN DATA WRAPPER</command>, <command>CREATE
    SERVER</command> and <command>CREATE FOREIGN TABLE</command> commands.
    The validator function must be registered as taking two arguments, a text
    array containing the options to be validated, and an OID representing the
    type of object the options are associated with (in the form of the OID
    of the system catalog the object would be stored in).  If no validator
    function is supplied, the options are not checked at object creation time.
   </para>

   <para>
    The foreign data wrappers included in the standard distribution are good
    references when trying to write your own.  Look into the
    <filename>contrib/file_fdw</> subdirectory of the source tree.
    The <xref linkend="sql-createforeigndatawrapper"> reference page also has
    some useful details.
   </para>

   <note>
    <para>
     The SQL standard specifies an interface for writing foreign data wrappers.
     However, PostgreSQL does not implement that API, because the effort to
     accommodate it into PostgreSQL would be large, and the standard API hasn't
     gained wide adoption anyway.
    </para>
   </note>

   <sect1 id="fdw-routines">
    <title>Foreign Data Wrapper Callback Routines</title>

    <para>
     The FDW handler function returns a palloc'd <structname>FdwRoutine</>
     struct containing pointers to the following callback functions:
    </para>

    <para>
<programlisting>
FdwPlan *
PlanForeignScan (Oid foreigntableid,
                 PlannerInfo *root,
                 RelOptInfo *baserel);
</programlisting>

     Plan a scan on a foreign table. This is called when a query is planned.
     <literal>foreigntableid</> is the <structname>pg_class</> OID of the
     foreign table.  <literal>root</> is the planner's global information
     about the query, and <literal>baserel</> is the planner's information
     about this table.
     The function must return a palloc'd struct that contains cost estimates,
     a string to show for this scan in <command>EXPLAIN</>, and any
     FDW-private information that is needed to execute the foreign scan at a
     later time.  (Note that the private information must be represented in
     a form that <function>copyObject</> knows how to copy.)
    </para>

    <para>
     The information in <literal>root</> and <literal>baserel</> can be used
     to reduce the amount of information that has to be fetched from the
     foreign table (and therefore reduce the cost estimate).
     <literal>baserel->baserestrictinfo</> is particularly interesting, as
     it contains restriction quals (<literal>WHERE</> clauses) that can be
     used to filter the rows to be fetched.  (The FDW is not required to
     enforce these quals, as the finished plan will recheck them anyway.)
     <literal>baserel->reltargetlist</> can be used to determine which
     columns need to be fetched.
    </para>

    <para>
<programlisting>
void
BeginForeignScan (ForeignScanState *node,
                  int eflags);
</programlisting>

     Begin executing a foreign scan. This is called during executor startup.
     It should perform any initialization needed before the scan can start.
     The <structname>ForeignScanState</> node has already been created, but
     its <structfield>fdw_private</> field is still NULL.  Information about
     the table to scan is accessible through the
     <structname>ForeignScanState</> node (in particular, from the underlying
     <structname>ForeignScan</> plan node, which contains a pointer to the
     <structname>FdwPlan</> structure returned by PlanForeignScan).  Note that
     this function is not called during an <command>EXPLAIN</> without the
     <literal>ANALYZE</> option.
    </para>

    <para>
<programlisting>
TupleTableSlot *
IterateForeignScan (ForeignScanState *node);
</programlisting>

     Fetch one row from the foreign source, returning it in a tuple table slot
     (the node's <structfield>ScanTupleSlot</> should be used for this
     purpose).  Return NULL if no more rows are available.  The tuple table
     slot infrastructure allows either a physical or virtual tuple to be
     returned; in most cases the latter choice is preferable from a
     performance standpoint.  Note that this is called in a short-lived memory
     context that will be reset between invocations.  Create a memory context
     in BeginForeignScan if you need longer-lived storage, or use the
     <structfield>es_query_cxt</> of the node's <structname>EState</>.
    </para>

    <para>
<programlisting>
void
ReScanForeignScan (ForeignScanState *node);
</programlisting>

     Restart the scan from the beginning.  Note that any parameters the
     scan depends on may have changed value, so the new scan does not
     necessarily return exactly the same rows.
    </para>

    <para>
<programlisting>
void
EndForeignScan (ForeignScanState *node);
</programlisting>

     End the scan and release resources.  It is normally not important
     to release palloc'd memory, but for example open files and connections
     to remote servers should be cleaned up.
    </para>

    <para>
     The <structname>FdwRoutine</> and <structname>FdwPlan</> struct types
     are declared in <filename>src/include/foreign/fdwapi.h</>, which see
     for additional details.
    </para>

   </sect1>

 </chapter>