README

Setup the database:
$ createdb liquid_feedback
$ psql -v ON_ERROR_STOP=1 -f core.sql liquid_feedback

Optionally insert demo data:
$ psql -v ON_ERROR_STOP=1 -f demo.sql liquid_feedback

Compile lf_update binary:
$ make

Ensure that "lf_update dbname=liquid_feedback",
"lf_update_issue_order dbname=liquid_feedback", and
"lf_update_suggestion_order dbname=liquid_feedback" are called
regularly! It is required to run these commands regularly
(i.e. every 5 minutes via cron) to update cached supporter counts,
to change the state of issues when neccessary, to calculate the
result of votings, etc.  If you wish, copy the created lf_update
and lf_update_suggestion_order files to /usr/local/bin or a
similar directory.

It is possible to run these two commands in parallel, if a setup
requires splitting the load to multiple processor cores. In other
cases it is recommended to run "lf_update" first, and then
"lf_update_issue_order" and "lf_update_suggestion_order".

On successful run, these commands will not produce any output
and exit with code 0. The commands "lf_update_issue_order" and
"lf_update_suggestion_order" may be called with a first argument
of "-v" to print log output.

NOTE: When writing to the database, some INSERTs must be executed
      within the same transaction, e.g. issues can't exist without
      an initiative and vice versa.

To create an export file, which is containing all but private data,
you may use the lf_export shell-script:
$ lf_export liquid_feedback export.sql.gz

Refer to source code of function "delete_private_data"() to see,
which data gets deleted. If you need a different behaviour, please
copy the function and modify the lf_export shell script accordingly.

To uninstall the software, delete the lf_update binary
and drop the database by entering the following command:
$ dropdb liquid_feedback

Updating is done using the update scripts in the update/ directory.
Do not forget to make backups, before installing any updates.
After updating it is recommended to recreate the schema as follows
(assuming the database superuser is named "postgres"):
$ pg_dump --disable-triggers --data-only DATABASE_NAME > tmp.sql
$ dropdb DATABASE_NAME
$ createdb DATABASE_NAME
$ psql -v ON_ERROR_STOP=1 -f core.sql DATABASE_NAME
$ su postgres  # (execute following command as database superuser)
$ psql -v ON_ERROR_STOP=1 -f tmp.sql DATABASE_NAME
$ exit         # leave "su" command
$ rm tmp.sql