On Thu, Jun 13, 2013 at 8:20 AM, Christopher Huhn, GSI <c.h...@gsi.de> wrote:
> So the documentation is completely wrong here: '-O file' has the same > timestamp of the last change of the web > page as the naked wget and behaves clearly different from '-O - > file' here. While it's true that -O file gets the timestamp (mainly because it can, whereas - generally can't), the text of the documentation is still exactly correct: the file is immediately truncated (opened O_TRUNC or "w" mode or what have you), before any timestamping checks are done rendering -N entirely useless in combination with it. Whether this is great design or not isn't really the issue (personally I kind of loathe the way -O was implemented, but didn't want to mess too much with history when I was maintaining it); it's the way it's been since the beginning. And the major point is that since it's opened in write mode, the timestamp will always be "now" at the moment it's opened. Giuseppe (current upstream maintainer) might entertain a bug report against that (or might not), but the documentation correctly describes the current behavior. It's also possible that newer wgets have already changed this behavior, though I kind of doubt it. To be clear, the test of the documentation's statement wouldn't be to check whether the file has a timestamp after downloading; it would be to check whether -N gets honored (refuses to download in the event that the timestamp of any existing file matches the remote). -mjc -- To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org
