Tasks
Stuff that needs doing. Tasks with no "owner" are open. To "claim" a task, edit this page and add your name as "owner", and email your thoughts to the amanda-hackers list. Other hackers will be happy to help. If you have other tasks, please add them here, following the pattern below. This forms a more in-the-moment version of the Wishlist/Features planned document.
The easiest way to get started on any of these projects is to fork Amanda and push your completed changes back.
Devices
Writing new backends for the Device API.
Optical Device
Peter Conrad and Richard Kunze wrote a patch for Amanda to write dumps to optical media, but it is written as a replacement for the taper, and doesn't work for modern Amandas. The better way to write to optical media is now to write a new device that handles on-disk caching of data and writing new volumes out to optical media as it is available.
Applications
Writing new Amanda frontends, using the Application API
Difficulty: | easy |
---|---|
Skills: | Perl, Postgres |
Owner: | User:Nikolas |
Postgres Application
Write an application to allow users to backup and restore Postgres databases.
MySQL Application
Write an application to allow users to backup and restore MySQL databases.
Cyrus imapd Application
Write an application to allow users to backup and restore emails in Cyrus imapd. While 'tar' can back up a mailspool, it is not ideal for the huge number of small files, and does not work well with so many hard-linked files.
AFS Application
Write an application to allow users to backup and restore from Andrew Filesystems; see How To:Backup AFS Volumes.
Authentication
OpenSSL auth
An OpenSSL-based security driver could enable encryption without the messy manual setup of SSH. More importantly, it can use certificate validation to perform bi-directional authentication of network communications, without reference to any UNIX usernames, passwords, or home directories.
Perlification
There is no shortage of stuff to rewrite in Perl. This is just a sampling of the "low-hanging fruit".
Update amtoc
amtoc is already in Perl, but it parses logfiles directly. Update it to be more readable, to use Amanda::DB::Catalog to read the logfile, and to produce correct output. It will need new unit tests.
Rewrite amlabel
amlabel is fairly simple: use the changer to load the correct tape, check for an existing label, and write a new label. It will need new unit tests.
Rewrite amtape
amtape is basically a user-friendly interface to the tape changer, and should be easy to rewrite in Perl. It will need new unit tests.
Build Process
More portable Perl extension building
The current build process uses the -I flags from perl to find perl.h, but then uses Libtool to do the final linking. This seems to work on most platforms, but there may be a better way. The problem with using MakeMaker, Module::Install or Module::Build is that they assume that the compiler which built perl is still available. This is not the case on platforms like Solaris. A "hybrid" solution -- which tests the native-perl method, and falls back to the current Libtool method -- may be best. We have yet to have any trouble building perl extensions, so this task may be moot.
Simplify Perl Automake rules
The Makefile.am stanza to build a perl extension from a SWIG input file is too long. Ideally, this could be expressed in a few lines giving only the source file and binary dependencies of the extension.
Translation
Amanda is (mostly) internationalized, and ready to have translations plugged in. Add a task here for each new locale.
I18n for Perl
To his shame, Dustin has been furiously writing un-internationalized Perl. This means that, at the moment, Perl applications won't respect any translations. Clearly, that should change!
German Translation
German is as good a first translation as any, and will work the kinks out of the translation process, paving the way for other locales.
Miscellaneous
Fix data encryption
Data encryption support in Amanda is not well-maintained. Some of the available scripts are Linux-specific, or fail in confusing ways. This task would involve a thorough exploration of the current support, documentation, and writing patches for any problems found in the existing code.