This HOWTO first explains the files in the scripts/pdb directory and then

describes common tasks involving pdb updating.

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

Files and directories in scripts/pdb:

HOWTO:

This howto document

basepath:

Directory containing the Fink package description files. This is a cvs

checkout of the 'fink' module but might be overwritten by rsync on the pdb

server when refreshing the pdb (see below for details)

basepath_fink.inc:

Config file with the path to the pdb script files (the current dir)

current_pdb_dump.sql:

SQL statements produced by the dump script containing the current contents of

the pdb.

drop.sql:

SQL statements to drop all pdb tables in the db. Only needed to restart from

scratch.

dump:

Shell script to dump the package descriptions. Used by the dumprelease and refresh

scripts.

dumprelease:

Shell script to dump the package description of a Fink released (defined by

a CVS tag). See below for a description on how to use it.

fink:

A directory containing the Fink tree with the package description files. This

will be updated by the refresh and dumprelease scripts

.finkdbi:

Config file with the information on how to connect to the database via perl

(Could alternatively be located in ~/.finkdbi)

finksql:

Shell script to connect to the database via the shell

fix_permissions.sh:

Shell script to set the file permissions

inject:

Shell script to inject the sql statements in current_pdb_dump.sql into the db

pdb_full_update-piecewise:

Shell script to update the pdb (calling the refresh and inject scripts)

distribution after distribution (10.2-gcc3.3 10.3 10.4 10.5)

This script is called every 8 hours on finch using the follwoing cron line:

20 1,9,17 * * * root nice /opt/pdb/pdb_full_update-piecewise 2>&1 | \

mail -s "Web-PDB Update `date`" fink-auto-logs@lists.sourceforge.net

pdb_full_update-unified:

Shell script to update the pdb (calling the refresh and inject scripts)

all in one step

refresh:

Shell script to refresh the package descriptions und dump them into

current_pdb_dump.sql. refresh is called by the pdb_full_update-* scripts

releases.sql:

SQL statements containing the information about current supported releases.

Needs to be updated whenever a release is added or no longer supported. See

below on how to use it.

*-stable.sql:

SQL statements containing the package information about Fink releases. Created

by the dumprelease script.

schema.sql:

SQL statements containing the database schema of the pdb. See below on how to

use it.

sections.sql:

SQL statements containing the information about the package sections. Needs to

be udpated whenever a new section is added. Called by the pdb_full_update-*

scripts.

update.sh:

Shell script to update the cvs directories and configure the local fink. See

below for more information.

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

Q: How do I update the package database?

A: You normally should never have to do this since we have a cron script in

place that performs that job every 8 hours for us. But if for some reasons it

is necessary to perform such an update right on the spot, proceed as follows.

First log into the finkproject.org server:

ssh MYUSERNAME@finkproject.org

The first time you are doing this, you have to create two auxiliary files:

~/bin/finksql (or ./finksql if there is no home folder):

#!/bin/sh

exec mysql -h localhost -u fink --password=DBPASSWORD fink

(or, on shell.sf.net: exec mysql -h mysql4-f.sourceforge.net -u f17203rw --password=DBPASSWORD f17203_pdb )

~/.finkdbi (or ./.finkdbi if there is no home folder):

DBI:mysql:database=fink;host=localhost|fink|DBPASSWORD

(or, on shell.sf.net: DBI:mysql:database=f17203_pdb;host=mysql4-f.sourceforge.net|f17203rw|DBPASSWORD )

You probably should chmod go-rw the two files, and also chmod u+x the finksql

script. If you don't know the value for "DBPASSWORD", ask the project admin.

Now change into the PDB directory:

cd /opt/pdb/

(or, on shell.sf.net: cd /home/groups/f/fi/fink/pdb )

Refresh the list of packages (essentially this runs a 'cvs up'):

./refresh

Finally, insert the package descriptions:

./inject

That should be it!

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

Q: How do I update the PDB for a full fink release?

A: Assuming you just released fink 0.4.5, you'll want to adjust the package

database accordingly.

First log into the finkproject.org server:

ssh MYUSERNAME@finkproject.org

We assume that you setup the necessary auxiliary files already, if not read

the above Q&A for information on how to do it.

Now change into the PDB directory:

cd /opt/pdb/

Next dump the list of packages in the release. Replace 0.8.1 as appropriate,

replace '10.4' with the Fink tree you want to count as stable, and replace

'i386' with the architecture you wish to use (i386 or powerpc):

./dumprelease 0.8.1 10.4 i386

(Or:

./dumprelease 0.8.1 10.4 powerpc

./dumprelease 0.7.2 10.3 powerpc

./dumprelease 0.6.4 10.2-gcc3.3 powerpc

)

This will produce a file called 0.8.1-i386-stable.sql. Next, you must update the

releases.sql file for 0.8.1.

You best do that in CVS using a developer checkout on your local machine,

then check the change in, and run "cvs up" on the shell server.

Once you did this, run these commands (order is important):

~/bin/finksql < releases.sql

~/bin/finksql < 0.8.1-i386-stable.sql

(If you have released 0.8.1 for multiple architectures, you can add a

command to this list such as ~/bin/finksql < 0.8.1-powerpc-stable.sql .)

That's it.

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

Q: How to set-up the pdb directory for the first time?

A: Do the following steps:

1)

Check out the 'scripts/pdb' module.

cvs -d:pserver:anonymous@fink.cvs.sourceforge.net:/cvsroot/fink login

cvs -z3 -d:pserver:anonymous@fink.cvs.sourceforge.net:/cvsroot/fink co -P scripts/pdb

2)

Check out the module 'fink' as the folder 'basepath' [sic] inside the just

checked out dir 'pdb'

cd scripts/pdb

cvs -z3 -d:pserver:anonymous@fink.cvs.sourceforge.net:/cvsroot/fink co -P -d basepath fink

3)

Create the following file:

- basepath_fink.inc: Add a line with the full path to the pdb script directory

E.g. '/opt/pdb'

(or, on shell.sf.net: '/home/groups/f/fi/fink/pdb' )

4)

Update the cvs directories and configure your local fink by doing:

./update.sh

5)

Set up the database by doing:

~/bin/finksql < schema.sql

(Use '~/bin/finksql < dump.sql' to clean out the db first)

6)

Initialize tables with:

~/bin/finksql < sections.sql

~/bin/finksql < releases.sql

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

The original content of this HOWTO was written by Max Horn, largely based on

some instructions given by Christoph Pfisterer. Use at your own risk.