> > > I'm hoping to run (exmh over) nmh over sshfs, but it doesn't work > > > because 1) sshfs doesn't implement link()... > > > > > > > When I ran Evolution a few years ago and told it about my MH message > > tree, it went through the tree and broke all of my message links. > > Because I like the ability to have a message appear in all of the > > folders where it might belong, I haven't used Evolution since. (I also > > filed a bug report, but I haven't checked whether anyone was able to fix > > it -- or did.) > > > > Still, for compatibility with non-*nix filesystems, would it makes sense > > for nmh to have a "no-links" configuration option? I haven't looked at > > the code to see how possible that might be... just an idea. > > Interesting idea. A one line hack made scan and show work over sshfs, > so I might do the work. > > Is there a regres/test suite/document for nmh???
FWIW: I just "audited" the link() calls w/ grep. With my previous patch, "CFLAGS=-DHAVE_LSTAT=0" and "|| linkerr == ENOSYS" added to line 182 of folder_addmsg.c, I now have a nmh build that (essentially) never calls link(). (There are still calls to link() in send.c and whatnowsbr.c, but they are used by dist which I don't use.) Running it over sshfs is of course a bit slower than just using ssh, but it allows me to run exmh (over broadband) faster. off to convince sendmail to relay via SMTP-AUTH over port 587... steve -- _______________________________________________ Nmh-workers mailing list [email protected] http://lists.nongnu.org/mailman/listinfo/nmh-workers
