WebDAV (http://www.webdav.org/) is simply
wonderful. I wonder if being able to "mount" a DAV
store from and HTTP server and then doing fileops
syncs, updates, etc. on it would be the right user
"metaphor" .. in some ways DAV would almost become
like NFS (!?)
It's true that a number of the recently "discovered" free projects were quite mature when they "made the news". But making the news in the mainstream media is hardly an accurate measure of the value or impact of a project though - remember much of the press seems to think ILOVEYOU.vbs is yet another example of cyberterrorism.
In a limited set of situations Koffice is already a competitor to Office2000 : it's twice as featureful as the commercial products were 5 years ago (remember Word6?) and it's free, easily extendable, writes to common XP formats, etc. (no NT or MacOS/X ports yet but ;-) but there's no hype or mainstream discussion yet because it's not on any radar screens. There surely will be astounding ways for databases, KOffice, GNOME, Gnumeric, CORBA, and DCOP to connect and cooperate that compare to anything on the market right now. Unfortunately the gradual boring process of maturing, adding features, minimizing bugs, and planning for the future development of applications like these doesn't capture media imagination the way benchmarks, market share, and other exciting skirmishes in the supposed "war" do.
I am astounded at the lack of attention that projects like Postgres get (contributing to database development requires fairly high level technical skills I presume). Postgres's gradual but steady improvement and the increasing power of cheap hardware has made it a real competitor in a whole host of implementations (Oracle, Sybase, etc. might be "better", but "good enough" and source available is sometimes more important). But this too has been a gradual development and hard to fit into the way the media packages reality and presents their analysis of technology developments.
There are projects that might be easier to contribute to but still have substantial impact: adding a scanner driver to SANE; extending USB support to more devices (keyboards, storage, audio, modems, scanners, network devices); NetBSD, FreeBSD, Linux people working together to create/maintain common interfaces so developing portable software is easier; contributing documentation to "desktop" projects like GNOME and KDE (esp. making it "fit" better with other non-Linux platforms); etc. etc. None of these are "flashy" visible projects of course.
Compared to the *BSDs an authoritative base set of high quality manual pages for Linux remains an "on going task": but keeping documentation up to date and in synch across all the I18N L10N environments is a massive task for *all* free OS projects. For Linux the LDP is a fantastic project in this regard - compared to dancing paper clips the LDP's Guides, manpages and HOWTO collection are a veritable treasure. It might not be high profile but it's crucial, valuable, open, and has as much potential for impact on OSS development and use in the future as another big software project might have.
There have been lots of important developments in the way commercial software companies relate to OSS too: Applix's opening up of ELF (nice), Metro-X contributions to XFree, Netware and OpenLDAP, that sort of thing. These are likely to be things that have long, slow, positive impacts - thus they may be difficult to fit into the media's "breathless relentless change" view of the universe.
Re: the above comments on automake autoconf - I didn't know the situation was quite that bad ;-) Then again there's a good chance that the whole range of tools might just get rewritten in python: