Hi,
While downloading a 12GB file, I noticed that wget (version 1.10.2)
crashed when trying to download it. After examining it a bit, I
discovered that the Apache server falsely sends the header
"Content-Length: -649674338" (the truncated 32-bit signed value of the
true length). When using --ignore-length the file is retrieved normally.
If the length given by the server is negative (i.e. surely bogus), wget
should automatically ignore it (with a possible warning message), not
crash.
Additionally it would be great if wget could correctly detect if the
content is actually larger than implied in the headers. If the truncated
file size is positive, wget would probably silently just download a piece
of the file (I haven't been able to test this in practise, but I'd guess
this to be the case). This could be done by attempting to download more
data than the Content-Length specifies, and see if any extra is available.
It may be questionable to save more data than specified in the headers,
but at any rate wget could print out a warning message suggesting the
usage of --ignore-length.
--
__________________________________________________
/____\ Sampo Niskanen <=> [EMAIL PROTECTED] \
\ http://www.iki.fi/sampo.niskanen/ \
\ ________________________________________\___
\___/___________________________________________/
