Project ideas

Things I'd like to see.

A "smart" link toolbar

Mozilla Firebird has a link toolbar, which uses the ancient <link rel=""> mark-up to provide easy access to the logical navigation of a multi-page document, and much more. But few pages actually use it. It would be neat to have an extension that browsed the page's DOM for links with content such as "Next" or "> Next", and added rel tags as appropriate. It could incorporate Ralph Churchill's Next/Prev Image, which works in a similar fashion by numerically incrementing filenames in the URL.

Update: LinkIt is an extension for Mozilla Firebird that does this.

Bibliographical database search library

EndNote is a pretty heavy-duty bibliography manager for Mac and Windows, with a heavy-duty cost attached to it. But really it's not doing much more than providing a handy interface to online bibliographical databases. It would be nice to see an extensible libbib. It even rhymes.

Generalised thesaurus library

Aiksaurus (available still from Abiword CVS) was an abortive attempt at such. OpenOffice has its own English thesaurus datasets, but the interface seems entwined within the OOo infrastructure. KOffice have a similarly entwined component called KThesaurus. It's high-time for a vendor-neutral libthesaurus all interested parties can use.

Update: I was pointed to Cryptic Muse. Maybe that's good enough.

Disassembler on steroids

Some distributions install binutil's libopcodes, but this is far away from what a truly useful platform-independent disassembly library would look like. It should support the standard bytes to asm instructions transformation, of course, but it would also support a programmatic interface for querying the assembly data. The ISA corpus would include information on classification of the instructions: you could ask, "How many forward branches rely on %eax ?", though that's perhaps a silly example. Extra credit if it provides a nice API for editing the binary too.

An IRC client that can render LaTeX

Seth Arnold and William Lee Irwin mentioned the idea (on IRC of course). You could use David Kastrup's excellent preview-latex to render the LaTeX snippets.

XML sanification wrapper

The XML tools (in particular xsltproc and friends) have nothing as simple and plainly obvious as gcc's -I (include file) and -D (define variable) options. These are particularly useful for any non-trivial use of auto-building docs. I complained about this a bit; amongst the useful tips I received, indicating that it's at least partially possible (as I mention in Using the GNU Autotools with DocBook XML and xsltproc), it was suggested that I make a wrapper script that hides the complexities of doing it the XML way. It would be a nice project. You could support -D by using XSLT parameters and the role="" attribute on <phrase>, and -I via XInclude.


Why can't I write Apache modules in HQ9+ ?


A nice home for all the pointless wasteful code in the world. We can start with the GNU C library's memfrob() and strfry() and build from there.

A CIA screensaver

Suggested by Isvara, it would be cool to see CIA commits hooked up to an xscreensaver module.

Traffic/bandwidth shaping GUI

It would be really nice to see a good GUI shell for traffic shaping.

ASCII Art dialog designer

There exist a number of ASCII art editors, but there are none specifically for user interface designers discussing designs via e-mail or in bug reports. A decent editor would work similarly to Qt's designer, allowing quick mock-ups using the standard ASCII art widget set.

Mutt and Maildir

Mutt has reasonable support for Maildir, but there's no real configuration possible for the folder view. You can't show how many new or existing mails there are in a folder (truly trivial with Maildir), nor can you only list maildirs that have new items only. Easy enough to do.

Update - I wrote a hacky patch (against 1.5.9) which more or less does what I want. It adds new %c and %n folder_format conversion specifiers, which list the total and new count of mails in the given folder respectively. It also filters out any dirs with no mails. I use it with the Maildir date-ordering patch linked from mutt.org.

RSS Aggregator

Although I've not done anything like an exhaustive search, most RSS readers I've tried seem to go out of their way to make things annoying for me. For example, Sage. Why on earth am *I* having to remember the last entry I read for each feed? Why on earth isn't it amalgamating all the feeds into one date-ordered list so I can just read "today's new entries?". And given how unsuccessful attempting to render the feed is, why can't it just go to the source site when I click on an RSS entry?

Anarchy in development, tyranny in integration. - Momchil Velikov