Return to TURBO.README CVS log

Up to [CSRG BSD Unix] / 43BSDReno / contrib / isode-beta / quipu

Annotation of 43BSDReno/contrib/isode-beta/quipu/TURBO.README, revision 1.1.1.1

1.1       root        1: This file documents the TURBO_DISK changes made by Tim Howes at the University
                      2: of Michigan.  
                      3: 
                      4: 1. Overview:
                      5: 
                      6: The TURBO_DISK option is designed to make updates to the quipu x.500
                      7: database much faster.  Normally when an entry is modified, quipu writes to
                      8: disk the entire edb file containing that entry.  If the edb file is at all
                      9: large, this can take quite some time.  The TURBO_DISK changes keep the disk
                     10: data in dbm files (actually gdbm files) instead of regular edb files.  This
                     11: way, when an update is made only the affected entry need be rewritten.
                     12: 
                     13: The performance increase you experience will be directly proportional to
                     14: the size of your edb files.  Whereas an update operation used to take time
                     15: proportional to the size of the edb file, with TURBO_DISK it takes constant
                     16: time.
                     17: 
                     18: 2. Using TURBO_DISK:
                     19: 
                     20: First, you'll need a copy of gdbm on your system.  Gdbm-1.3 or later is
                     21: recommended.  Second, you'll need to make one small change to the file
                     22: turbo_load.c, if you are using the TURBO_LOAD option.  The line:
                     23: 
                     24: #include "/u/up/tim/src/gdbm-1.3/gdbmdefs.h"
                     25: 
                     26: should be changed to reflect the actual location of the gdbmdefs.h
                     27: include file.  Note that this file is not installed in the normal process
                     28: of installing gdbm.  Disgustingly enough, the turbo_start routine needs
                     29: to be able to stat the gdbm file, so it needs the defs in this file.
                     30: If you are not using the TURBO_LOAD option, this step should not be
                     31: necessary.
                     32: 
                     33: Using the TURBO_DISK option involves two steps.  First, you must compile
                     34: the dsap library and quipu with the TURBO_DISK option defined.  The easiest
                     35: way to do this is to put the following line in ./h/config.h:
                     36: 
                     37: #define TURBO_DISK     /* enable fast edb update operations */
                     38: 
                     39: You will then need to add
                     40: 
                     41:        -lgdbm
                     42: 
                     43: to the "LSOCKET =" line in config/CONFIG.make.
                     44: 
                     45: Then make quipu as usual.  The second step involves converting the edb
                     46: file hierarchy into a gdbm file hierarchy.  The ./quipu/tools directory
                     47: has some tools to do this for you.  The shell script tree2dbm is provided
                     48: for this purpose.  Use it like this:
                     49: 
                     50:        tree2dbm database-directory
                     51: 
                     52: where database-directory is the directory where the edb file hierarchy
                     53: begins.  This script does a find on that directory for files named EDB
                     54: and runs them through the edb2dbm program which creates a file called
                     55: EDB.gdbm.  Alternatively, you can run edb2dbm by hand on each EDB
                     56: file.  The original EDB file is neither removed nor molested, so
                     57: you'll need twice the disk space.  The up side is that your old edb
                     58: files are still around in case you want to back off to the non-turbo
                     59: quipu.
                     60: 
                     61: 3. Other things to note:
                     62: 
                     63: With the TURBO_DISK option parse errors will be reported somewhat differently
                     64: in the dsap.log file.  Since line numbers don't make much sense in a 
                     65: gdbm file, errors will be reported based on the key of the offending
                     66: entry.  If you get a parse error (because of a non-printable character,
                     67: for example), the best approach is to do something like this:
                     68:        edbcat EDB >bob
                     69:        now edit bob and fix the problem
                     70:        edb2dbm bob
                     71:        mv bob.gdbm EDB.gdbm
                     72: where this procedure assumes you are in the directory containing the
                     73: bad EDB file.  The edbcat program can be found in the ./quipu/tools
                     74: directory and is used to convert from gdbm back to plain text format.

This archive runs on limited infrastructure. Preserving old code on modern bandwidth. Automated agents are requested to crawl responsibly.