The other thing more or less is ripped from the Windows DL-Manager
FlashGet (but why not). Wouldn't it be useful if wget retrieves a file
to a temporary renamed filename, for instance with the extension .wg! or
something and renamed back to the original name after finishing? Two
TL advantages IMO: First you can easily see at which point
a download broke (so you don't have to look for a file by date or size
or something in a whole lot of them).
The other is the possibility to resume a broken download with the
option -nc (so the already downloaded files aren't looked up again).
Wget needn't check a lot and could determine by the file extension
that this is the one file where it has to continue.
TL wget needs to remember a LOT more than simply the last file that was being
TL downloaded. It needs to remember all the files it has looked at, the files
TL that have been downloaded, the files that are in the queue to be downloaded,
TL the command line and .wgetrc options, etc.
TL With some clever planning by someone who knows the internals of the program
TL really well, it might be possible for wget to create a resumption file with
TL the state of the download, but I'm guessing that is a huge task.
Well, I said I don't know what it takes and if it makes sense
programming-wise. And actually I thought it wasn't about wget getting
to remember more. If it creates a resumption file then it
no-clobbers all the complete downloads (no remembering) when the
broken download has to be repeated, doesn't
find the current incompleted one (because of the extension), starts to download (again
with
resumption extension), finds there is one when it tries write and decides to
continue for that file at the right point.
Well, the conventional way of finding the broken file, deleting it and
start again with -nc works too, of course. :-)
-- Brix
