On Mon, 17 Sep 2001, Ethan Benson wrote:
> On Mon, Sep 17, 2001 at 09:21:36PM +1000, ivanr@xxxxxxx wrote:
> > On Mon, 17 Sep 2001, Ethan Benson wrote:
> > > > If you'd rather fix the source, check cmd/xfsdump/librmt/rmtopen.c line
> > > > 80.
> > >
> > > actually that code looks fine, its the _rmt_dev function that is
> > > broken, here is a patch:
> > Well, I probably would've removed the call to rmt_dev, as Tim did
> > did -- no point having a function called rmt_dev if it doesn't have
> > anything to do with devices or /dev.
> well it looked to me like rmt_dev was what was figuring out whether
> the pathname given was a remotehost:/path or a local pathname. but
> again i am not familier with this code at all.
Yes, it does do that ... but it only does it by checking to see if there's
a ':' in the name. You'll note that Tim simply replaced the call to
rmt_dev with a call to strchr. There really isn't much point in keeping
rmt_dev if it's only function is to call strchr, and if it has nothing to
do with checking whether the path is a device.
But still, we're quibbling over details -- your main point that xfsdump
shouldn't be restricted to remote devices seems valid and unless we
discover an unexpected problem, we're happy to remove that restriction.
(Although, I must admit, Tim was the one who wanted to change it and I
originally wanted to keep the status quo ... :)