 |
 |
Hi Fons, In xrootd there is a "forbidden" character set which also includes '?'. Generally, the forbidden characters are those that might allow someone to escape the shell should the filename be passed by an external method. While some people will argue that this is not generalized. I would submit that this is an academic argument that does not stand on practical grounds. We aren't implementing a full generalized filesystem but a data access system intended for people who generally have no such requirements. Andy ----- Original Message ----- From: "Fons Rademakers" <[log in to unmask]> To: "Fabrizio Furano" <[log in to unmask]> Cc: "Andrew Hanushevsky" <[log in to unmask]>; "Andreas Joachim Peters" <[log in to unmask]>; "Derek Feichtinger" <[log in to unmask]>; "xrootd mailing list" <[log in to unmask]> Sent: Wednesday, April 27, 2005 6:22 AM Subject: Re: olbd fails to learn when a file disappears from a leaf node, but another copy still exists > Do we have to worry about supporting filenames containing '?' ? Support > this via '\?' to escape the '?'. Might be useful in the posix layer since > we cannot assume anything about a name. > > -- Fons > > > Fabrizio Furano wrote: >> Hi, >> >> I argue from this that the opaque info passed e.g. through xrdcp must be >> passed for any request containing a filename, like Stat or Dirlist. Am I >> right? >> >> Fabrizio >> >> Andrew Hanushevsky wrote: >> >>> Hi Andreas, >>> >>> OK, so it would appear that we will need to extract out the information >>> after the "?" and pass that as a separate parameter. I do that, instead >>> of >>> passing the complete url, so as to not re-implement searching for the >>> opaque information in every function. The called function, hoewver, is >>> responsible for making sense of the opaque information. >>> >>> That does mean changing most file system calls to include the opaque >>> parameter. That also solves the olbd issue in a unified way. >>> >>> Do we all agree to go that route? >>> >>> Andy >>> >>> On Tue, 26 Apr 2005, Andreas Joachim Peters wrote: >>> >>> >>>> As it is, it is absolutely fine for me. I would prefer, that the >>>> complete >>>> URL is always passed to any function and the function has to extract >>>> the part >>>> it needs. But as it is, it works perfectly for us. >>>> >>>> I use the following syntax: >>>> >>>> root://server.domain:port/<lfn>?&authz=<authorization block> >>>> >>>> Because even for a stat command it can be useful, that you can specify >>>> some environment variable like the stagepool the file is on. >>>> >>>> Cheers Andreas. >>>> >>>> > > -- > Org: CERN, European Laboratory for Particle Physics. > Mail: 1211 Geneve 23, Switzerland > E-Mail: [log in to unmask] Phone: +41 22 7679248 > WWW: http://www.rademakers.org/fons/ Fax: +41 22 7679480 >