postgresql/INSTALL

POSTGRESQL INSTALLATION INSTRUCTIONS
Copyright (c) 1996 Regents of the University of California

This directory contains the source and documentation for PostgreSQL
(version 1.09) PostgreSQL is a derivative of POSTGRES 4.2 (the last
release of the UC Berkeley research project).  For copyright terms for
PostgreSQL, please see the file named COPYRIGHT.  This version was
developed by a team of developers on the postgres developers mailing
list.  Version 1 (through 1.01) was developed by Jolly Chen and Andrew
Yu.


REQUIREMENTS TO RUN POSTGRESQL
------------------------------

PostgreSQL has been tested on the following platforms:

	alpha		-	DEC Alpha AXP on OSF/1 2.0
	hpux		-	HP PA-RISC on HP-UX 9.0
	i386_solaris	-	i386 Solaris
	sparc_solaris	-	SUN SPARC on Solaris 2.4
	sparc		-	SUN SPARC on SunOS 4.1.3
	ultrix4		-	DEC MIPS on Ultrix 4.4
	linux		-	Intel x86 on Linux 1.2 (or above) ELF or a.out
	BSD44_derived	-	OSs derived from 4.4-lite BSD (NetBSD, FreeBSD)
        bsdi            -       BSD/OS 2.0 and 2.01
        bsdi_2_1        -       BSD/OS 2.1
	aix		-	IBM on AIX 3.2.5
	irix5		-	SGI MIPS on IRIX 5.3 
	dgux            -       DG/UX 5.4R3.10
  Some hooks are provided for
	svr4		-	Intel x86 on Intel SVR4
	next		-	Motorola MC68K or Intel x86 on NeXTSTEP 3.2
  but these are guaranteed not to work as of yet.

PostgreSQL is also known to work on a number of other platforms that the
authors have not personally tested.

You should have at least 8 MB of memory and at least 30 MB of disk space to
hold the source, binaries, and user databases.


MIGRATING FROM POSTGRES VERSION 1.0
-----------------------------------

Version 1.01 and 1.02 (and above) are mostly backward compatible with Version
1.0, but the database format is incompatible, so if you have databases that
you use with Version 1, you need to convert them before you can use them with
Version 1.02.  Once you do that, you won't be able to use them with Version 1
anymore.

For details on how to do this conversion, see the files doc/MIGRATION_1.0_to_1.01
and MIGRATION_to_1.02.1


INSTALLING POSTGRESQL
---------------------

Installing PostgreSQL encompasses only installing the software on your system
so you can use it to access (or create or manipulate) databases.  This
step does not include actually creating any database or configuring your 
system to use it.

To install PostgreSQL on UNIX platforms:

1. Unpack the source distribution into a source directory.  We'll assume
   "/usr/src/postgres95" in this discussion.  This should be a new directory.
  
2. Set your current directory to the source directory:

   cd /usr/src/postgres95

3. Build PostgreSQL:

   If you're installing PostgreSQL on Ultrix 4.x or Linux, see the 
   porting notes at the end for additional packages that you need to install
   before installing PostgreSQL.

   If using Linux or Irix, you should also read the machine-specific FAQs.

   Our Makefiles require GNU make (called gmake in this document) and
   also assume that "install" accepts BSD options. The INSTALL
   variable in the Makefiles is set to the BSD-compatible version of
   install. On some systems, you will have to find a BSD-compatible
   install to the location of this program. (eg. bsdinst, which comes
   with the MIT X Window System distribution) 

   Customization can be done by editing src/Makefile.global. You may change
   the various configuration options here, such as where the PostgreSQL
   executable files are installed and where postgres looks for the database
   directory.  The configuration switches are fairly self-explanatory, but we
   will go over some of the more commonly-changed options:

     - PORTNAME specifies the platform on which PostgreSQL is being built.
       This is set to UNDEFINED. You will need to change it to reflect
       your platform. (sparc for SunOS 4.1.x, sparc_solaris for Solaris
       2.4, ultrix4 for Ultrix 4.4, and hpux for HP-UX 9.0, etc.)

     - SRCDIR specifies where the source files are located. (defaults to
       $(POSTGRESDIR)/src.)

     - POSTGRESDIR specifies the top-level directory where PostgreSQL
       binaries, header files, libraries, and databases are installed.

     - NAMEDATALEN and OIDNAMELEN allows you to set the maximum length of
       system identifiers (table names, function names, etc.)  It
       defaults to 32.  You may alter this if you like, but be aware that
       databases created with different NAMEDATALEN's do not
       interoperate.

     - USE_READLINE specifies whether you want to use the GNU readline and
       history libraries for the psql interactive frontend program.  GNU
       readline is not supplied with PostgreSQL and can be found in the
       usual ftp sites for GNU software.

     - HBA specifies whether you wish to use host-based authentication
       for PostgreSQL.  See the section "How to Create a Database System"
       for how to set up the HBA permissions if you decide to use HBA.

   After editing src/Makefile.global, you are ready to compile PostgreSQL
   (it takes about 10 minutes on a 133Mhz Pentium running linux):

     % cd src
     % gmake 

   The gmake ultimately issues the message "All of Postgres95 is
   successfully made.  Ready to install."  If you don't get that, the make
   failed, and there should be error messages at the end detailing why.

4. Install PostgreSQL

   Installing just means placing all the files built in the previous step
   into their live locations on your system. 

     % gmake install

   This will narrate all the files being installed.  You should watch and 
   be sure the files are going to reasonable places and confirm for yourself
   that they ended up where they belong.

   Any error messages indicate something is wrong and you probably have to
   correct it before PostgreSQL will work.


HOW TO CREATE A DATABASE SYSTEM
-------------------------------

Once you have Postgres installed, you'll need at least one database system
on which to operate.  A database system is a collection of databases that
are used together and fall under a single authority.  You can have as many
database systems as you want on a single unix system.

You select a unix user to be the "postgres superuser" for a database
system and that user, for one thing, owns all the unix files that hold
all the data for that database system.  It is usually a good idea to create
a user for the sole purpose of being a postgres superuser.

WARNING: PostgreSQL is not secure.  Anyone who can connect to a database
system can easily assume all the unix privileges of its Postgres
superuser.  The simplest way is by creating and running a C language
function.  There are plans to remedy this in future developent.

The program initdb (part of Postgres) is what initializes (creates) a
database system.  Initdb uses the defaults specified in Makefile.global.
See the man page for initdb for more information.

  % initdb

By default, the user issuing the initdb command becomes the Postgres
superuser, and only the unix superuser can specify any other user as the
Postgres superuser.

Setting up Permissions
----------------------

The first thing you should do after creating a database system is set up
the permissions for connecting to the database.  These are kept in the 
file pg_hba in the data directory.  Initdb creates a sample version of
this file, which contains comments telling you how to set it up.

The Postmaster Daemon
---------------------

Finally, in order to use the database system, you'll need to have a
postmaster daemon running.  There is one postmaster process per database
system.  The postmaster runs the program "postgres" and must run as the
Postgres superuser.  See the postgres man page.

So, for example, you can login as the Postgres superuser and issue the
command:

  % postmaster -S -D/usr/lib/postgres/postgres_data -p5432

This says to run the postmaster against the database system created above,
to accept connections from users on the conventional TCP port 5432, and
(-S) to run in the background without issuing messages about normal 
execution.

This is a good daemon to start via system startup scripts, using su (be
careful NOT to run the postmaster as the unix superuser by mistake).


TESTING POSTGRESQL
------------------

We suggest you run the regression tests to make sure the release was
installed successfully and works as designed in your environment.  The
regression tests can be found in src/test/regress. (see
src/test/regress/README for more details)

     % cd /usr/src/postgres95/src/test/regress
     % gmake all runtest

This will run a whole slew of regression tests and might take a long time
to run.  When it's done, the output is in the file obj/regress.out.  You
can compare this to a sample run that we supply in the file
sample.regress.out. (You should get roughly the same output except for
some pathnames.)

     % diff obj/regress.out sample.regress.out

The regression test takes about half an hour to run on a Sparc 10.  You
may want to use 'grep -v' to remove unsignificant differences.


PLAYING WITH POSTGRESQL
-----------------------

After PostgreSQL is installed, a database system is created, a postmaster
daemon is running, and the regression tests have passed, you'll want to 
see PostgreSQL do something.  That's easy.  Invoke the interactive interface
to PostgreSQL, psql, and start typing SQL:

  % psql -p 5432 template1

(psql has to open a particular database, but at this point the only one
that exists is the template1 database, which always exists.  We will connect
to it only long enough to create another one and switch to it).

Note that we have told psql to connect to Port 5432, which is what we told
the postmaster to listen on when we started it above.

The response from psql is:

  type \? for help on slash commands
  type \q to quit
  type \g or terminate with semicolon to execute query
You are currently connected to the database: template1

template1=> 

Create the database foo:

template1=> CREATE DATABASE FOO;
INSERT 773248

(Don't ever forget those SQL semicolons.  Psql won't execute anything until it
sees the semicolon).

template1=> \c foo
closing connection to database: template1
connecting to new database: foo

(\ commands aren't SQL, so no semicolon.  Use \? to see all the \ commands).

template1=> CREATE TABLE bar (column1 int4, column2 char16);
CREATE

template1=> \d bar

...

You get the idea.



QUESTIONS? BUGS? FEEDBACK?
--------------------------

First, please read the Frequently Asked Questions and answers in the file
called FAQ.

If you still have questions, please send them to
postgres95@postgres95.vnet.net.

If you have a bug report to make, please send a filled out version of
the file named "bug.template" to pg95-dev@ki.net.

If you would like to help out with the development and maintenance of
PostgreSQL, send subscribe to the developers mailing list.  See
README.support for more information

----------------------------------------------------------------------

Porting Notes:
-------------
Ultrix4.x:
	You need to install the libdl-1.1 package since Ultrix 4.x doesn't
	have a dynamic loader. It's available in
	   s2k-ftp.CS.Berkeley.EDU:pub/personal/andrew/libdl-1.1.tar.Z

Linux:
	The linux port defaults to the ELF binary format. (Note that if you're
	using ELF, you don't need dld because you'll be using the dl library
	that comes with Linux ELF instead.)

	To compile on non-ELF Linux, comment out the LINUX_ELF line in
	src/mk/port/postgres.mk.linux. Also, the dld library MUST be obtained
	and installed on the system. It enables dynamic link loading capability
	to the postgres port. The dld library can be obtained from the sunsite
	linux distributions. The current name is dld-3.2.5.
				(Jalon Q. Zimmerman 
				<sneaker@powergrid.electriciti.com> 5/11/95)

	To compile with flex, you need a recent version (2.5.2 or
	later). Otherwise, you will get a 'yy_flush_buffer' undefined error.
        Note, however, that flex v2.5.3 has a bug. See the FAQs.

BSD/OS:
	For BSD/OS 2.0 and 2.01, you will need to get flex version 2.5.2
	as well as the GNU dld library.  Flex version 2.5.3 has a known bug.

NeXT: 
	The NeXT port was supplied by Tom R. Hageman <tom@basil.icce.rug.nl>.
	It requires a SysV IPC emulation library and header files for 
        shared libary and semaphore stuff.   Tom just happens to sell such 
        a product so contact him for information.  He has also indicated that
        binary releases of PostgreSQL for NEXTSTEP will be made available to
        the general public.  Contact Info@RnA.nl for information.