Limit dataset name size to 256 characters to prevent segfaults if the checkpoint

Limit dataset name size to 256 characters to prevent segfaults if the checkpoint
files has very many datasets. This is just a dirty hack...

darcs-hash:20040320151749-19929-18b22f38e48a294374aa5b5c664a22ca7109253c.gz

1 files changed, 4 insertions, 3 deletions

diff --git a/Carpet/CarpetIOHDF5/src/iohdf5chckpt_recover.cc b/Carpet/CarpetIOHDF5/src/iohdf5chckpt_recover.cc
index a11955fa9..498d26138 100644
--- a/Carpet/CarpetIOHDF5/src/iohdf5chckpt_recover.cc
+++ b/Carpet/CarpetIOHDF5/src/iohdf5chckpt_recover.cc
@@ -18,7 +18,7 @@
 #include "cctk_Version.h"
 
 extern "C" {
-  static const char* rcsid = "$Header: /home/eschnett/C/carpet/Carpet/Carpet/CarpetIOHDF5/src/iohdf5chckpt_recover.cc,v 1.12 2004/03/20 15:43:37 cott Exp $";
+  static const char* rcsid = "$Header: /home/eschnett/C/carpet/Carpet/Carpet/CarpetIOHDF5/src/iohdf5chckpt_recover.cc,v 1.13 2004/03/20 16:17:49 cott Exp $";
   CCTK_FILEVERSION(Carpet_CarpetIOHDF5_iohdf5chckpt_recover_cc);
 }
 
@@ -243,7 +243,7 @@ namespace CarpetIOHDF5 {
     int currdataset,ndatasets;
     char * name;
 
-    char datasetname[1024];
+    char datasetname[256];
 
     int varindex; 
 
@@ -259,7 +259,7 @@ namespace CarpetIOHDF5 {
     MPI_Bcast (&ndatasets, 1, MPI_INT, 0, dist::comm);
     assert (ndatasets>=0);
 
-    char datasetnames[ndatasets+1][1024];
+    char datasetnames[ndatasets+1][256];
 
     for (currdataset=0;currdataset<ndatasets+1;currdataset++){
       if (myproc==0) {
@@ -267,6 +267,7 @@ namespace CarpetIOHDF5 {
       } //myproc = 0
     }
 
+    
     if (h5verbose) cout << "ndatasets: " << ndatasets << endl;
 
      for (currdataset=0;currdataset < ndatasets+1;currdataset++) {