 |
 |
Hi Gregory, I committed a fix in the head that should meet the needs of your code. Now you can read starting beyond EOF, or crossing it, and you get the correct number of bytes. Maybe Pete wants to make a new release out of it. I put also a small test program useful to see what happens requesting byte blocks in sequence. A typical use of it is: TestXrdClient_read root://noric04.slac.stanford.edu//tmp/fabrizio_grossofile 131072 15000000 2 and then you have to type couples (len, offset) to perform the actual reads. Or just start it without pars to get the help. Anyway, please note that the caching does not guarantee that you will gain performance. In some cases you can get a true boost, but in other cases things could also go worse than without. It depends on the parameters, yes, but also on what the application is doing, and on the performances of the machines, network, etc. If you are interested in performances, IMHO the only way to get an answer is to try. Thank you very much for your patience, please let me know. Fabrizio Gregory J. Sharp wrote: > > On 26Jul2005, at 15:48 , Fabrizio Furano wrote: > >> ok, at least it's working. >> >> When I am satisfied with the next fix (I hope tomorrow afternoon >> Italian time), please may I ask you to redo your test without those >> lines? > > > Sure. No problem. It only takes 20 seconds to recompile and a few > seconds to run it. > > -- > Gregory J. Sharp email: [log in to unmask] > Wilson Synchrotron Laboratory url: http://www.lepp.cornell.edu/ > ~gregor > Cornell University ph: +1 607 255 4882 > Ithaca, NY 14853 fax: +1 607 255 8062 >