diff options
author | Bill Wendling <wendling@ncsa.uiuc.edu> | 2003-03-19 23:39:14 (GMT) |
---|---|---|
committer | Bill Wendling <wendling@ncsa.uiuc.edu> | 2003-03-19 23:39:14 (GMT) |
commit | 43eb81fb3c6997ded3a94593d543c413dfa2ce5a (patch) | |
tree | 6fd24a6afad6b69a30ea9677214d0ee2f6a4ea13 /src/H5Fistore.c | |
parent | 0e83643f8f6505b9fb23989e1286f36572e34f72 (diff) | |
download | hdf5-43eb81fb3c6997ded3a94593d543c413dfa2ce5a.zip hdf5-43eb81fb3c6997ded3a94593d543c413dfa2ce5a.tar.gz hdf5-43eb81fb3c6997ded3a94593d543c413dfa2ce5a.tar.bz2 |
[svn-r6504] Purpose:
Why not?
Description:
Lots of changes. This is basically a "commit because there are a lot
of changes" commit.
The server now acts more like a metadata caching server. The dumping
mechanism finally works (I think). At least in the test I did it
seemed to write things back. There's a new MPI TAG to use
(H5FP_TAG_DUMP) when dumping things. Changed a lot of the functions
from passing a structure around to just passing a pointer to that
structure. I have no clue why I did it the previous way...
Um...And more synchronization stuff between the client and server.
The protocol between the two was being broke in a couple of places.
We're getting closer!
Platforms tested:
Linux (probably Modi4 too)
Misc. update:
Diffstat (limited to 'src/H5Fistore.c')
0 files changed, 0 insertions, 0 deletions