Thanks Andy, and I think this simple solution solves xrdcat and xrdcp problem, so we can have only one. Artem. On Thu, 16 Sep 2004, Andrew Hanushevsky wrote: > I would adopt url syntax to generalize specifying data sources and targets. > The use of "cat" is Unix legacy in the sense that cp never had a way to > specifying that the target was stdout. So the gurus hard linked cat to cp so > that the program would know, by the name it was invoked, how to configure > the output. > > For pipes, I propose we use: > > pipe:[//<hostname>[/<pipe application>]] > > The default pipe application, if not specified, would be stdout for output > and stdin for input. The default host would be localhost. In fact, for > xrdcp, it would make no sense at this point to allow anything but localhost > for the pipe location. > > For instance, using xrdcp to take data from application "foobar" and feed it > into an xrootd server would look likeL > > xrdcp pipe://localhost/foobar xrootd://xrd-rdr03/store/file.root > > You can imagine sematically going wild with this syantx, but for now, we > should keep it real simple. > > Andy > > > ----- Original Message ----- > From: "Peter Elmer" <[log in to unmask]> > To: "Artem Trunov" <[log in to unmask]> > Cc: "Fabrizio Furano" <[log in to unmask]>; > <[log in to unmask]> > Sent: Thursday, September 16, 2004 9:48 AM > Subject: Re: requirements for xrdcp > > > > On Thu, Sep 16, 2004 at 09:39:35AM -0700, Artem Trunov wrote: > > > > Artem, you want to read the log files remotely at all costs, don't > you? > > > > > > I realized some time ago, that the cost is your dead body... > > > > Ah, was my response at some point "over my dead body"? I don't think > > I ever said that about reading the log files. Usually that response is > > evoked when someone starts talking about contacting databases from worker > > nodes and other stupid things like that. > > > > An 'xrdcat' could in fact be doable. ;-) Then one just have to organize > > the logs so that they are exported, which I already wanted to do to > harvest > > them for other reasons. And I guess it could allow you to deal with the > > HPSS/staging logs or whatever else on the machine, too. > > > > Pete > > > > ------------------------------------------------------------------------- > > Peter Elmer E-mail: [log in to unmask] Phone: +41 (22) 767-4644 > > Address: CERN Division PPE, Bat. 32 2C-14, CH-1211 Geneva 23, Switzerland > > ------------------------------------------------------------------------- > > > >