Prompted by a comment on jwz's question about using a flash as a focusing aid for shots in the dark, I started looking into Astro Photography and how they do focusing.
The most directly useful hints, are hints on camera settings and manual focus, particularly:
ensure eyepiece diaopter setting is correct (something which I need to find a good way to verify, as I think it's only vaguely right for my eyes)
focus on the brightest thing you can find (especially for stars)
check focus electronically by taking a test shot and using review zoom (10x, compared with typical negative review loupe at 5x)
add magnification, eg, Canon Angle Finder C which in addition to allowing adjustments of eyepiece viewing angle also includes 2.5x zoom on eyepiece (apparently the Olympus Varimagni Finder will also fit on (some?) Canons)
the half-shutter-press can also confirm focus spots with lens in manual focus
(if supported) live video feed out to a computer (which, alas, the EOS 400D doesn't support to the best of my knowledge; the EOS 450D and EOS500D do support liveview; some people have hacked liveview together with a webcam)
(There are also various suggestions which are more appropriate for bright sources -- such as stars -- against dark backgrounds, which given suitable defraction mask or Hartmann mask helps with focusing. The Hartmann mask in particular is in focus when the multiple images converge. There is also a newer Bahtinov mask design which is apparently more useful for astrophotography.)
This astrophotography focusing guide contains lots of hints, some specific to astrophotography and some more reusable. Amongst the most useful hint is that many auto-focus lenses will focus "beyond infinity", ie that there's a point beyond infinity being in focus where it infinity again out of focus, so that the last point on the lens may not be the one you want.
Historically cameras had ground glass (difusing) focusing screens, often with a split prism or microprism ring, which split the image such that the parts of the image converged only when they were in focus. But ubiquitous autofocus makes this these manual focusing aids much less common, which in turn makes manual focus more difficult. (Amongst other issues the focusing screens would consume quite a bit of the light, which caused issues with focusing on dark items or smaller apetures -- anything smaller than f/5.6 seems problematic, and f/2.8 was traditional for good focusing, which appears in part to be an artifact of the need for large apetures for good deflection of out of focus light. This, combined with the relatively dark viewfinders in a lot of cheaper DSLRs probably contributed to removing the focusing aids in the viewfinder when autofocus was used almost exclusively.)
There were lots of different types of focusing screens for different purposes. This page at focusingscreen.com has lots of good examples, and they appear to sell a range of replacement focusing screens (and provide install instructions, eg EOS 400D focusing screen install, EOS 450D focusing screen install instructions). Katz Eye Optics also have a range of replacement focusing screens for Canon cameras, as do Haoda (review.
FTR, this is by far the best writeup of how both split prism manual focus and phase comparision auto focus works (cached copy) that I found. (Also AF accuracy specifications in Canon Cameras by the same author; DPReview posting linking to it which has a good summary; Cambridge In Colour tutorial on autofocus, and Wikipedia page; and difference between phase comparision and contrast detection autofocus; contrast detection is slower to perform, but doesn't require the SLR mirror to be down and requires no pre-calibration, so is what is offered when LiveView is in use; a lot of compact cameras apparently also use contrast detection, since it can be implemented entirely in software. And note that phase comparision works in only one dimension, so for higher accuracy two sensors are needed one horizontal and one vertical; in most entry to midrange cameras only the central autofocus central is both horizontal and vertical, and typically the other sensors are vertical sensors so focus best on a horizontal line.)
Also of interest:
Testing focus (on consumer cameras the best guarantee is that the focus point will be within the depth of field, but not always the thing most in focus)
details of close-up focusing rings which screw onto the filter mount, and macro lenses
manual button and serial port shutter release control (and other articles on DSLRs by same author, including image correction curves andthe need for a linear conversion when doing noise reduction, dust issues and photoshop hints for aerial images through windows)
CHDK script for multiple long exposures (CHDK is a third party firmware enhancement for certain Canon cameras, including my Canon IXUS 80, which I keep meaning to play with.)
According to the instructions to obtain the version number, I have firmware 1.01A on my IXUS 80 which seems suitable -- note that the "set" button must be pressed first, closely followed by the "disp" button; holding down the "set" button gives what appears to be an easter egg showing the time in rotating cubes, and the date if the camera is vertical rather than horizontal.
CHDK Av Tv and ISO overrides, and high speed flash sync (because many compact digital cameras have electronic shutters they have very high speed flash sync, unlike many DSLRs with focal plane shutters; see also explanation for mechanical shutters -- amongst other reasons the circuit real estate needed for electronic shutter reduces the light gathering ability), custom alignment grids, and RAW shots (conversion tool to DNG; source on Google code, which has Mac and Linux versions, and there is a FreeBSD port)
EOS Documentation Project, which mostly consists of links to external documentation at this point, but looks to be a useful index of relevant PDFs
Finally those astrophotographers that wanted more red and IR light into their camera resorted to extreme surgery on their cameras to replace the IR filter with plain glass. (Links to more hardware modifications.)
And now for something completely different: Google Scholar is a Google hosted index of US case law, similar to WestLaw, but gratis; and snapshot.debian.net provides point-in-time packages from Debian which is useful for retrieving arbitrary older versions of packages to check them.
ETA, 2010-04-04: Photo.Net post on how Auto Focusing works, which amongst other things explains that the need for a larger aperture (geometry of bending of light, rather than the amount of light per se; at too small an aperture the amount of bending of light from the two edges is too small to be reliably detected, hence most cameras cannot auto-focus at f/8 or smaller).
And there is a cool high speed video of mirror flip up and flip down which very clearly shows the mirror flap and the resulting vibrations.
ETA, 2010-07-11: Lens Rentals guide to Autofocus