Re: feature request: split --fetch to --post-queued --fetch-overview --fetch-selected-bodies
Mirko Liss <[email protected]> Tue, 11 Feb 2003 04:20:18 +0100
| Newsgroups | gmane.network.noffle.user |
|---|---|
| Message-ID | <[email protected]> |
Miernik wrote: [..fetch less, fetch faster..] > I think that would be very simple for someone who knows Noffle > internals. Please look at noffle/src/noffle.c, at the procedures doFetch() and main(). It is _very_ easy to implement, indeed. Just copy doFetch() and delete some of the 'connOk = connOk &&'-lines. The command line options will be parsed using a switch()-case-structure in the main()-procedure. It's easy to insert a new option. However, I'm not convinced that such a patch is actually useful. This way, you can't avoid downloading the overviews, you'll just delay it. What do you think about a fetch mode that does actually avoid downloading overviews of new articles? You won't be able to filter articles and you won't get articles from groups that have been subscribed in 'over' or in 'thread' mode. Instead of getting overviews with 'XOVER number-number', noffle sends 'XHDR Message-Id number-number' to find out the Message-Ids of the new articles. That's still a waste of bandwidth, but I suppose it's a bit more efficient than getting the whole overviews. Well, if noffle was able to download articles by number instead of by Msg-Id, the aforementioned 'XOVER' could be skipped without the replacement 'XHDR Message-Id'. That would be even faster. But it's a lot of work to code this, and there are some problems without trivial solutions. Avoiding multiple download of crossposts, for example. regards, Mirko ------------------------------------------------------- This SF.NET email is sponsored by: SourceForge Enterprise Edition + IBM + LinuxWorld = Something 2 See! http://www.vasoftware.com