I wrote my own transfer script to bring the files from the SD card to the computer (I configured it as the default action to take when a new card is detected in my card reader).
It copies the files from the card in a directory named "~/photos/
yyyy/
yyyymmddnn/", where
nn is a sequential number starting at 01, useful in case of multiple transfers on the same day.
The files are renamed as they are transfered, to
yyyymmddnn-xxx where
yyyymmddnn is the same as the directory name, and
xxx is a sequential number. I shoot mostly RAW+JPG these days, and both files are kept with the same name, but the relevant extension (i.e. .pef, .dng, or .jpg).
The files are
not automatically removed from the card. I manually format each card after I have confirmed the transfer is complete.
The whole hierarchy is hosted on a mirrored filesystem, and I also rsync the whole thing to two other machines.
I browse the images with Gthumb (who cheerfully ignores the RAW files), deleting as I go along. This only deletes the JPG, however, so I have written a "cleanup" script to delete "orphan" RAW files (i.e. those with no corresponding JPG file) periodically.
I have linked the "0" key on the keypad in Gthumb with a script that calls Gimp with either the JPG being viewed or its corresponding RAW version if the latter exists. RAWs are handled by the UFRaw Gimp plugin.
I use Gimp for corrections (rotation, shadows&highlight, b&w conversions, etc), sharpening (Eg-SmartSharpen), and resizing. I have written a Gimp script to automate the resize-for-the-web-sharpen-and-add-a-border operation.
I usually try to keep an "original" JPG version under the initial filename. I use "original" in quotes because this will often be the RAW-converted file, sometimes with cropping or other non-trivial modification. Other files will get suffixes appended to their filename, ex: "2009071001-005-crop-bw-web.jpg" would be a cropped, black & white, "web" version of the image 20090710-005.jpg.
Well, it makes sense to me anyway