werc

Werc is a minimalist content management system (CMS) with a web interface implemented using the rc shell and the Plan 9 from Bell Labs tools provided as part of Plan 9 from User Space.

Originally developed for cat-v.org it has become more generalized. One of its major features is that it allows management of multiple domains and virtual hosts easily from a single installation.

(More or less idiosyncratic) Features

• Purely file system based, all documents are stored in a file tree as plain text files (.html files and .txt files are also handled and integrated transparently), no need for any database.
• Also allows for dynamically generated pages/scripts/applications that can take advantage of the framework (for an example see the dynamically generated gsoc project status table).
• Easily embedding of pre-existing html documents into the framework.
• Easy management of multiple domains/sub-domains/virtual-hosts from a single installation (just add another directory under sites/), and easy sharing of templates and files among sites.
• Markdown templating plus template cache (thanks Mechiel!) to make up for dog-slow markdown implementations.
• Automatic sitemap generation.
• Bloging system including rss feeds.
• More that I probably have forgotten about.

Development

To get a copy of the latest development code using mercurial, do:

hg clone http://gsoc.cat-v.org/hg/werc/

You can also browse the online repository.

Dependencies

To run werc you will need a recent version of Plan 9 from User Space and a sane version of awk (ideally the original one maintained by Brian Kernighan). Compiling statically p9p and awk is probably a good idea and will improve performance on platforms with dog slow fork(2) (linux fork(2) is a few magnitude orders faster for statically linked binaries).

If you use the included markdown.pl you will need perl too, sorry.

TODO

Too much to list here.

• Blog:
  • Atom feeds
  • Comments (including threading)
  • Web interface to submit new posts
  • Some form of pagination/history browsing
  • Better modularization/integration of blog/rss related code.
• Sitemaps:
  • In the sitemap index pages should be used for description of directories
  • Dynamic generation of sitemap.txt too
  • Cache sitemap generation
  • Dirfilter var has different semantics in sitemap filtering than in sidebar filtering(?)
• Layout improvements:
  • Fixed width of left sidebar (so design scales down to cellphone-sized-screens)
  • Links to search page in top-right toolbar
  • Need to find a way to rig the order of items in the sidebar
• Better page titles (include whole path hierarchy?)
• Allow utf-8 characters in path elements (need to make sure this is safe)
• General code:
  • Write a regression test suite
  • Replace all reference to non p9p/p9 programs
  • Reorganize execution flow to allow for header changes imediately before page generation(useful to set redirections, caches, cookies, and errors)
  • Better defined 'API' for sub-apps (both a set of environment vars apps can rely on (and in some cases set) and functions they can call)

Future plans:

• Wiki mode (replacement for diri)
• Bug/issue tracking module
• Further modularization and extensibility (meta-templates)
• A way to wrap external applications (eg., hg/git browser)
• A way to generate static sites (ie., use werc as a pure templating system)
• 9P interface
• New (simplified) markdown implementation, ideally in C or awk (maybe smu/libsmu by gottox?):
  • Simplified
  • Tables
  • Images
  • Autogenerated anchors
  • No inline html

Resolved issues and implemented features:

• Permanent redirections by setting a variable in _config, avoiding need to mess around with httpd configuration for page moves. (done, set redirectPermanent var)
• Broken links in blog titles in post listing, eg: http://repo.cat-v.org/libixp/news/ (fixed?)
• Too wide margins for .html (troffied) in low screen resolutions (fixed)
• Better display of single blog entries (done, maybe could use more work)
• Make a proper 0.1 release (done)
• Autogenerated sitemaps (both .txt and .html versions) (done, but probably needs more work/caching and only html version so far)
• Modularize footer the same way the top header is (done)
• Should set <BASE href=...> header so links and images in 'implicit dirs' work even when there is no trailing /. (Resolved in a better way by redirecting always to the canonical dir path ending with /)

Known bugs

• http://gsoc.cat-v.org/blog/asdfasdf/blog/ should give a 'not found' error
• Give proper 404 errors for non existent pages (1.0 BLOCKER)
• Listing of files/dirs with (single) dots is broken (1.0 BLOCKER)
• Some cat-v.org bits still left over here and there, should make them configurable (1.0 BLOCKER)

Contact

For patches and bugrepports email: uriel99-gmail.com

License

Public domain (by default includes markdown.pl which has a different license, but this can be replaced).

Credits

Thanks to Kris Maglione (aka JG) for implementing rss feeds, for writing the awk rc-templating system, and other help and inspiration (some parts of the code were based on JG's diri wiki).

Thanks to Mechiel (aka oksel) for the md_cache script.

Thanks Garbeam (aka arg) for writing the original diri code and showing that writing complex web apps in rc was feasible.

And thanks to everyone else whom I have forgotten and that has provided fixes and feedback.

---