Re: [HACKERS] pg_dump --comment? - Mailing list pgsql-patches

From Bruce Momjian
Subject Re: [HACKERS] pg_dump --comment?
Date
Msg-id 200406050457.i554vg312811@candle.pha.pa.us
Whole thread Raw
In response to Re: [HACKERS] pg_dump --comment?  (Tom Lane <tgl@sss.pgh.pa.us>)
List pgsql-patches
Tom Lane wrote:
> Bruce Momjian <pgman@candle.pha.pa.us> writes:
> > Tom Lane wrote:
> >> What happens in a pg_dump -Fc / pg_restore scenario?
>
> > So, in non-ascii format, I need to dump a comment record and read it
> > back on restore?  Yuck.  Anyone want to tackle that?
>
> Yes.  Hacking the output routines alone is almost never the right way
> to do things in pg_dump.
>
> My feeling is that if anyone cares to tackle this, it would be
> appropriate to add a record type that carries a user-supplied
> comment, and then people could do something like
>     pg_dump --comment "generated on `date`"
> if they want to have a timestamp.  I don't want a timestamp put
> into the file by default, because that creates problems for
> comparing dump files.

OK, new patch.  This puts start/stop times into dumps if you use -v
during dumping.  It creates a new node type:

    --
    -- PostgreSQL database dump
    --

    SET client_encoding = 'SQL_ASCII';
    SET check_function_bodies = false;

    --
    -- Name: DUMP TIMESTAMP; Type: DUMP TIMESTAMP; Schema: -; Owner:
    --

    -- Started on 2004-06-05 00:41:25 EDT

    ...

    --
    -- Name: DUMP TIMESTAMP; Type: DUMP TIMESTAMP; Schema: -; Owner:
    --

    -- Completed on 2004-06-05 00:44:28 EDT


    --
    -- PostgreSQL database dump complete
    --

One issue is that the system dumps certain settings into the output file
in ascii dumps, but pg_restore sets them on its own during the restore.
Because the timestamp can't be done during pg_restore, it has to be a
special node.

--
  Bruce Momjian                        |  http://candle.pha.pa.us
  pgman@candle.pha.pa.us               |  (610) 359-1001
  +  If your life is a hard drive,     |  13 Roberts Road
  +  Christ can be your backup.        |  Newtown Square, Pennsylvania 19073
Index: doc/src/sgml/ref/pg_dump.sgml
===================================================================
RCS file: /cvsroot/pgsql-server/doc/src/sgml/ref/pg_dump.sgml,v
retrieving revision 1.70
diff -c -c -r1.70 pg_dump.sgml
*** doc/src/sgml/ref/pg_dump.sgml    31 May 2004 13:37:52 -0000    1.70
--- doc/src/sgml/ref/pg_dump.sgml    5 Jun 2004 04:22:15 -0000
***************
*** 403,409 ****
         <para>
      Specifies verbose mode.  This will cause
      <application>pg_dump</application> to output detailed object
!         comments in the dump file, and progress messages to standard error.
         </para>
        </listitem>
       </varlistentry>
--- 403,410 ----
         <para>
      Specifies verbose mode.  This will cause
      <application>pg_dump</application> to output detailed object
!         comments in the dump file, start and stop times, and progress
!         messages to standard error.
         </para>
        </listitem>
       </varlistentry>
Index: doc/src/sgml/ref/pg_dumpall.sgml
===================================================================
RCS file: /cvsroot/pgsql-server/doc/src/sgml/ref/pg_dumpall.sgml,v
retrieving revision 1.43
diff -c -c -r1.43 pg_dumpall.sgml
*** doc/src/sgml/ref/pg_dumpall.sgml    29 Nov 2003 19:51:39 -0000    1.43
--- doc/src/sgml/ref/pg_dumpall.sgml    5 Jun 2004 04:22:15 -0000
***************
*** 192,199 ****
        <listitem>
         <para>
      Specifies verbose mode.  This will cause
!     <application>pg_dumpall</application> to print progress
!     messages to standard error.
         </para>
        </listitem>
       </varlistentry>
--- 192,200 ----
        <listitem>
         <para>
      Specifies verbose mode.  This will cause
!     <application>pg_dumpall</application> to output start and stop
!         times in the dump file, and progress messages to standard error.
!         It will also enable verbose output in <application>pg_dump</>.
         </para>
        </listitem>
       </varlistentry>
Index: src/bin/pg_dump/pg_dump.c
===================================================================
RCS file: /cvsroot/pgsql-server/src/bin/pg_dump/pg_dump.c,v
retrieving revision 1.373
diff -c -c -r1.373 pg_dump.c
*** src/bin/pg_dump/pg_dump.c    3 Jun 2004 00:07:36 -0000    1.373
--- src/bin/pg_dump/pg_dump.c    5 Jun 2004 04:22:21 -0000
***************
*** 32,37 ****
--- 32,38 ----
  #ifdef HAVE_TERMIOS_H
  #include <termios.h>
  #endif
+ #include <time.h>

  #ifndef HAVE_STRDUP
  #include "strdup.h"
***************
*** 163,168 ****
--- 164,170 ----
  static const char *fmtQualifiedId(const char *schema, const char *id);
  static int    dumpBlobs(Archive *AH, void *arg);
  static void dumpDatabase(Archive *AH);
+ static void dumpTimestamp(Archive *AH, char *msg);
  static void dumpEncoding(Archive *AH);
  static const char *getAttrName(int attrnum, TableInfo *tblInfo);
  static const char *fmtCopyColumnList(const TableInfo *ti);
***************
*** 598,603 ****
--- 600,608 ----
       * in a safe order.
       */

+     if (g_fout->verbose)
+         dumpTimestamp(g_fout, "Started on");
+
      /* First the special encoding entry. */
      dumpEncoding(g_fout);

***************
*** 615,620 ****
--- 620,628 ----
          dumpDumpableObject(g_fout, dobjs[i]);
      }

+     if (g_fout->verbose)
+         dumpTimestamp(g_fout, "Completed on");
+
      /*
       * And finally we can do the actual output.
       */
***************
*** 1280,1285 ****
--- 1288,1322 ----
      destroyPQExpBuffer(dbQry);
      destroyPQExpBuffer(delQry);
      destroyPQExpBuffer(creaQry);
+ }
+
+
+ /*
+  * dumpTimestamp
+  */
+ static void
+ dumpTimestamp(Archive *AH, char *msg)
+ {
+     char buf[256];
+     time_t now = time(NULL);
+
+     if (strftime(buf, 256, "%Y-%m-%d %H:%M:%S %Z", localtime(&now)) != 0)
+      {
+         PQExpBuffer qry = createPQExpBuffer();
+
+         appendPQExpBuffer(qry, "-- ");
+         appendPQExpBuffer(qry, msg);
+         appendPQExpBuffer(qry, " ");
+         appendPQExpBuffer(qry, buf);
+         appendPQExpBuffer(qry, "\n");
+
+         ArchiveEntry(AH, nilCatalogId, createDumpId(),
+                      "DUMP TIMESTAMP", NULL, "",
+                      false, "DUMP TIMESTAMP", qry->data, "", NULL,
+                      NULL, 0,
+                      NULL, NULL);
+         destroyPQExpBuffer(qry);
+     }
  }



pgsql-patches by date:

Previous
From: Tom Lane
Date:
Subject: Re: [HACKERS] pg_dump --comment?
Next
From: "Laurent Ballester"
Date:
Subject: Re: eventlog fix