On Tue, 9 Apr 2002 [EMAIL PROTECTED] wrote: > > Richard Sharpe writes: > > > OK, I have made a change to the 2.2.x code stream that tries again after > > *SMBSERVER fails, if the server was passed in as an IP address. > > > > Not tested, only built! > > > > Of course, this will be slow, as it will try to connect with the IP > > address as the called name first, then *SMBSERVER, and finally will do a > > name_status request (which takes time) and tries to connect with that > > name as the called name. > > > > Hmm... that sounds OK, but I already implemented something like your first > idea from last night. :) I should have written back to you sooner. > > My crawler has a couple of stages to it, and by the time it reaches the > full-out share crawling stage it already knows everyone's hostname that > it's going to crawl. So, I just added a static char * to libsmbclient that > holds the name of the current server, and a modifier to change it. Then, > in opendir and stat (the only two functions I'm really using) I set server > = the static string and voila, everything's A-OK... it's still incredibly > speedy, too. > > I don't know if something like this would be useful enough to actually > merge into libsmbclient... even so, it'd take some work to make it not so > quick-hack like. :) Let me know what you think.
There are a whole bunch of changes that I have made to allow me to control a whole bunch of things, and your change is just another one of these fine-tuning tweaks that seems to be needed. What I need to do is find a clean way to support this sort of stuff. Regards ----- Richard Sharpe, [EMAIL PROTECTED], [EMAIL PROTECTED], [EMAIL PROTECTED]
