Vladimir Volovich <[EMAIL PROTECTED]> writes: > Hrvoje> The inconsistency is a bug. It is intended that Wget encodes > Hrvoje> all the unsafe characters, both in files and directories. > Hrvoje> (It is debatable whether that is a bug.) This patch makes it > Hrvoje> consistent, but I will not apply it before the 1.8.1 release: > > i did not look at the result of the patch, but does it result in > ENCODING of all URL-unsafe characters when saving the file on the > local filesystem?
It does. (As implied by the "It is intended" part of the paragraph.) > i think wget should ENCODE all unsafe characters when making a > request to HTTP or FTP server, but wgen SHOULD NOT encode URL-unsafe > characters when saving files on local filesystem, because they are > not unsafe for the filesystem. Maybe you're right. Still, the fact is that *some* characters are unsafe in files -- for example, a literal \0 and the '/' character. Also, 8-bit stuff is not too safe to have in a file name. Then there are the spaces, backslashes, quotes, and directories that begin with tilde. Encoding file name characters the same as URL characters seemed like a neat idea several years ago, but maybe you're right in that it should be rethought.
