basic information
name: David Harris
e-mail: [EMAIL PROTECTED]
homepage: http://www.davideous.com/
preferred cpan user-id: DHARRIS
description of what I'm planning to contribute:
Right now I have a DB_File::Lock module that I'm ready to contribute and have
packaged up. I've also created a number of other modules for my job which may
be generally useful. Here are two examples of things I might contribute,
depending on the interest:
FreezeThawLight -- my own light implementation of the FreezeThaw module which
is twice as fast, which I created because FreezeThaw was just too slow and
feature heavy for me.
Pigeon::Server, Pigeon::Client -- library for creating a generic pre-forking
multi-process server which will handle connections and
generic requests for those connections, all with appropriate timeouts and
concurrency limits. On top of this module I built a RPC system for my mod_perl
fronted to call a slave backend which switches uid/gid and does stuff as user
accounts.
description of what I've got to contribute now
Right now I have the following module to contribute...
Name DSLI Description Info
------------- ---- -------------------------------------------- -----
DB_File::Lock RdpO DB_File wrapper with flock-based locking blank
(I'm not sure if the interface style part of the DSLI should be "O" or left
blank. This seems purely "object oriented", but the DB_File module DSLI lists a
blank interface style.)
The module is a wrapper for DB_File which properly adds locking. This makes
DB_File usable in environments such as mod_perl where multiple accesses to the
same file are expected. This also properly unlocks the database on object
destruction , which is a must for mod_perl environments.
A few months ago, I discovered a problem in the recommended locking procedure
in the DB_File POD documentation and a small example locking wrapper in the
mod_perl guide which was based on that code. I created a test harness and sent
it to Stas Bekman (mod_perl guide maintainer) and Paul Marquess (DB_File
author), who both verified the bug. In this bug report, I included a reference
to a simple one-file proof-of-idea locking wrapper module I created called
DB_Wrap. Paul changed his DB_File documentation on locking to refer to my
simple module and asked me to submit it to CPAN so he could properly refer to
it. Stas modified my module and inserted it in his mod_perl guide and is also
waiting on it to be in CPAN.
The bug report and my DB_Wrap module were also kicked out to the perl5-porters
list and discussed here. People didn't have too much to say and no problems
were found.
Since then, it has taken me a while to get around to formally packaging my
module up. I've packaged it up using the name DB_File::Lock and added in a few
features. Paul Marquess has seen reviewed a copy of the module in it's final
form.
Another note: I asked Paul if he wanted to merge this into his DB_File
distribution, and he said he didn't. He'd rather have me maintaining a separate
module.
BTW, this module is different than the current Tie::DB_Lock in its goals and
locking method.
that's it, right?
I think that covers everything. I hope I didn't provide too much detail.
- David Harris
Principal Engineer, DRH Internet Services
