XROOTD-L Archives

Support use of xrootd by HEP experiments

XROOTD-L@LISTSERV.SLAC.STANFORD.EDU

Options: Use Forum View

Use Monospaced Font
Show Text Part by Default
Show All Mail Headers

Message: [<< First] [< Prev] [Next >] [Last >>]
Topic: [<< First] [< Prev] [Next >] [Last >>]
Author: [<< First] [< Prev] [Next >] [Last >>]

Print Reply
Subject:
From:
Gerardo Ganis <[log in to unmask]>
Date:
18 Aug 2004 19:08:39 +0200Wed, 18 Aug 2004 19:08:39 +0200
Content-Type:
text/plain
Parts/Attachments:
text/plain (104 lines)
  Hi Fons, 

  Of course, the final aim should be at a complete build of
  the client on all the platforms.

  A separate target to make only the XrdSec&C shared libs
  can be implemented quickly: the porting of these modules 
  (and XrdOuc, used by them) to Win32 will surely take longer,
  since there are a few things to be adapted ...

  I think that modifying the way the client uses XrdSec, so that 
  it does not require those libraries at build time, is anyhow 
  a good thing, because it reduces the minimal client dependence
  on the server to the availibility of a few header files. 
  It should also (potentially) allow to build the client on a larger 
  number of platforms already from now. 

  Then we could concentrate on the porting of XrdSec and XrdOuc to
  Win32.

  What do you think?
  
  Cheers, Gerri    


On 18 Aug 2004, Fons Rademakers wrote:

> Yes, the absolute priority should be to port xrootd to all platforms
> except for Win32 (rootd also was never ported to Win32 either). However,
> the client definitely should work on Win32 (although some strong
> authentication modules might be missing if the supporting libs are not
> available on win32). So if we could include a special target in the
> xrootd Makefile for the authentication modules that can build on any
> platform, incl Win32, we are on the right way.
> 
> -- Fons
> 
> 
> On Wed, 2004-08-18 at 14:28, Peter Elmer wrote:
> >   Hi Fons,
> > 
> >   I think had planned to port the full xrootd build to all of the relevant 
> > unix platforms, but, like rootd, weren't planning on porting it to Win32. 
> > Perhaps we can at least make the security modules build on Win32, though, 
> > which would at least allow for the full functionality of the client. (One
> > could imagine some build option for xrootd which just build the subset on
> > Win32, which should be easier than the full port.) Andy, what do you think?
> > 
> >                                    Pete
> > 
> > On Wed, Aug 18, 2004 at 02:21:54PM +0200, Fons Rademakers wrote:
> > > On Wed, 2004-08-18 at 12:18, Gerardo Ganis wrote:
> > > > > On Tue, Aug 17, 2004 at 12:10:16PM +0200, Fons Rademakers wrote:
> > > > > > The main issues I've at the moment are:
> > > > > > 
> > > > > > - build procedure of sec modules separate from xrootd.
> > > > > >   Currently we can build xrootd and therefore the sec modules only
> > > > > >   on Linux and Solaris, however we really require the netx client to be
> > > > > >   able to build on all ROOT supported platforms, incl Win32. Since for
> > > > > >   the time being the building of the client is linked to the building
> > > > > >   of the server we have the client only on Linux and Solaris.
> > > > > 
> > > > >   Gerri and I talked about this yesterday. He thought there were two 
> > > > > problems:
> > > > >  
> > > > >   o Something in the way the packages are built within ROOT which made the
> > > > >     netx client build dependent on having built xrootd. Do you have some
> > > > >     mechanism in the ROOT builds which just skips rootd/xrootd on platforms
> > > > >     where it is not supported (e.g. Win32)?
> > > > > 
> > > 
> > > Currently they are dependent since the authentication modules are part
> > > of the xrootd tar ball and are build by the xrootd makefile. These
> > > authentication modules should be independent from xrootd and be ported
> > > to all possible platforms (ideally xrootd should be ported to all
> > > platforms so that this issue becomes moot).
> > > 
> > > 
> > > > >   o There was an explicit dependency on libXrdSec.so. It seemed like it
> > > > >     should be possible to make this load instead as a plugin.
> > > > 
> > > >   This is almost ready.
> > > > 
> > > > > 
> > > > >   Hmm, one thing occurs to me: how is the security stuff handled for Win32
> > > > > if the xrootd server isn't built? 
> > > > 
> > > >   I am modifying things in such a way that, when the server cannot be built,
> > > >   the protocol header files are extracted from the server source, so that 
> > > >   they can be found when building the client. 
> > > >   Of course, since lib/libXrdSec.so is not there, the client will not able 
> > > >   to satisfy strong authentication requests from the server; however it should
> > > >   work if no authentication or host authentication are required.
> > > > 
> > > I rather have a solution where the authentication modules are ported to
> > > all platforms and delivered as a separate tar. However, porting xrootd
> > > to all platforms should be the priority and will solve this problem even
> > > better.
> > 
> > 
> > 



ATOM RSS1 RSS2