Hi Rene, On Wed, Dec 24, 2003 at 11:44:15AM +0100, Rene Brun wrote: > Did you write your Tree with 3.10/01 or an old version? I wrote the tree with ROOT 3.10/01. > In 3.10/02, I added a fix when reading very old files. > Could you try with the released version 3.10/02 or tell me where > I can take one of you files (small file if possible)? I've put an example file here: /afs/cern.ch/user/e/elmer/public/root/convert5.02E.root The tree demonstrating the problem is "esd" and the branch (in this example) with the problem is "Emc_Bumps__n". This tree has 8000 events, just enough to get to the point where the relevant buffers is compressed. When I run with fewer events (7000) and the buffer is not yet compressed, I see that the contents are basically all zeroes with a very small number of "1"'s. A standlone test of filling a branch with 0's and 1's doesn't lead to the same problem. I'll setup and try reading with ROOT 3.10/02 even if this is not an old file... thanks, Pete > On Tue, 23 > Dec > 2003, Peter Elmer wrote: > > > Hi, > > > > Starting recently (probably from ROOT 3.10-01) we have started to see > > R__unzip errors in reading back tree branches containing only integers. It > > is a bit complicated so I've not yet succeeded in reducing it to a simple > > test case, but I already see something strange in looking at a file: > > > > *............................................................................* > > *Br 19 :Emc_Bumps__n : Emc_Bumps__n/i * > > *Entries : 6000 : Total Size= 36446 bytes One basket in memory * > > *Baskets : 0 : Basket Size= 32000 bytes Compression= 1.00 * > > *............................................................................* > > *Br 21 :Emc_Clusters__n : Emc_Clusters__n/i * > > *Entries : 6000 : Total Size= 36464 bytes One basket in memory * > > *Baskets : 0 : Basket Size= 32000 bytes Compression= 1.00 * > > *............................................................................* > > *Br 23 :Emc_Cands__n : Emc_Cands__n/i * > > *Entries : 6000 : Total Size= 36446 bytes One basket in memory * > > *Baskets : 0 : Basket Size= 32000 bytes Compression= 1.00 * > > *............................................................................* > > > > Is it normal that the total size would exceed the basket size yet a basket > > is not compressed and written separately from the tree? > > > > I recall that you put in some bug fix recently related to compression of > > branches containing integers, could you remind me what was that fix? > > > > [For the record, we are working on Linux (RH7.2) with gcc 2.95.3.] > > > > thanks, > > Pete ------------------------------------------------------------------------- Peter Elmer E-mail: Peter.Elmer@cern.ch Phone: +41 (22) 767-4644 Address: CERN Division PPE, Bat. 32 2C-14, CH-1211 Geneva 23, Switzerland -------------------------------------------------------------------------
This archive was generated by hypermail 2b29 : Sun Jan 02 2005 - 05:50:05 MET