Robert A. Schmied kindly contributed the following text (as well as a minor
correction) concerning use of beancounter on non-Linux / GNU systems,
i.e. those whose set of common Unix utilities may differ slightly.
Non-GNU Installation Notes for beancounter
(especially applicable for Solaris 2.x systems)
Systems that don't have bash at /bin/sh and other non GNU utils may find the
following guidance helpful. Such systems include SUN Solaris 2.x, and others
that derive from the AT&T SVR4 UNIX baseline and many non-Linux systems.
The general installation sequence remains unchanged.
00. Read all the information files: README, beancounter.html
0. Install PostgreSQL
1. Configure PostgreSQL (create user(s) and allow TCP/IP db connections)
2. Start postgres database server daemon
3. Build, test and install all beancounter dependent perl modules
( Date::Manip(5.35), DBI(1.16), DBD::Pg(0.93), HTML::Parser(2.20),
HTTP::Request(1.23), LWP::UserAgent(1.62), Statistics::Descriptive(2.4),
Text::ParseWords(3.1), Finance::YahooQuote, Data::Dumper, Getopt::Long )
4. perl -w Makefile.PL
5. setup_beancounter # performs the initial db creation and config
6. update_beancounter # performs required updates for new db schema
7. beancounter ready for your use, re-read usage instructions in
beancounter.html and BeanCounter.html
NB: step 6. is REQUIRED even on an initial beancounter installation
(yes, it should be called automatically by setup_beancounter at step 5).
If, at step 5, you get message like:
setup_beancounter: syntax error at line 37: `progname=$' unexpected
You aren't running a sh that understands bash syntax. You will need to
get and install bash (on your own) or use the korn shell (ksh) that ships
with SUN Solaris 2.x.
To use ksh, the first line of each file setup_beancounter and
update_beancounter should be changed to: #!/bin/ksh -e
If, at step 5, you get the database error message like:
*** Error: No postgresql user 'ras'
We were unable to start psql as the user 'ras' does not exist
You need to create a Postgresql user 'ras' first:
Change to user postgres: $ su - postgres
Create the user: # createuser ras -d -A
Exit from user postgres: # exit
and then run this script again.
You need to use the following postgresql create user command:
/path-to-pgsql-bin/createuser ras -d -A
If you get an error from createuser like:
createuser: invalid option: -d
You are encountering the command line parameter 'out of order'
problem. Please, send an email to Dirk about this if only to note the
fact that this condition does occur in environments other than mine ;-).
Please indicate your OS and shell in the message.
Re-ordering the arguments as follows will solve it:
/path-to-pgsql-bin/createuser -d -A ras
NB: this situation will likely occur in the setup_beancounter and
update_beancounter scripts as well, you can either make the changes
listed below now, or fix each error one step at a time ...
Then repeat step 5.
If you continue to get the database error message above and are sure
the user id is valid and the database daemon is running or if, at step
5, you get an error from psql like:
psql: FATAL: user "-l" does not exist
psql: FATAL: user "-c" does not exist
psql: FATAL: user "-q" does not exist
You are encountering the parameter 'out of order' problem within
the script. Both scripts will need changes as follows:
For script setup_beancounter (3 places):
change all occurrences of sequences like '$DBCOMMAND dbname -f'
sequences like '$DBCOMMAND -f dbname'
where -f is the (all if multiple) flag option (-l, -q, ...)
dbname are parameters without a leading '-' (template1, beancounter, ...)
There are 3 occurrences including the help message described above.
For script update_beancounter, the correction is a bit different:
Change the line:
query="select distinct symbol from stockinfo where active"
query="\t select distinct symbol from stockinfo where active"
also change the line:
echo $query | $DBCOMMAND -t | grep -q "\b[[:digit:]]"
echo $query | $DBCOMMAND | grep -q "\b[[:digit:]]"
This change is using the psql \t meta-command instead of the psql -t
command line argument.
Consider sending an email to Dirk about your need to do this if only
to note the fact that this condition does occur in environments other
than mine ;-). Please indicate your OS and shell in the message.
If, when running update_beancounter at step 6, you get a message like:
grep: illegal option -- q
Usage: grep -hblcnsviw pattern file . . .
You are using a non POSIX compliant grep. A GNU grep will work, as
will the SUN Solaris grep at /usr/xpg4/bin. The SUN Solaris grep at
/usr/5bin will not work.
If, after giving it some personal sweat equity, you run into problem(s)
that you are unable to resolve by yourself contact Dirk or myself, we
will try to help you resolve them. Dirk is the key for anything that
relates to beancounter, including the database. I, on the other hand
can assist you with script portability issues on non-GNU systems such
as Solaris and other ATT based UNIX systems.
The assistance pleading message should include: the command line and
all resulting output plus as much information about your environment,
operating system and as much other info you deem appropriate for remote
problem analysis and diagnosis.
ps -- constructive comments regarding this text, including variations
due to other operating systems and the like are welcome. We prefer
diffs created via 'diff -ub orig change' (but will accept diff -cb for
those Solaris users that don't have the GNU diff) -- ras