Currently implemented in batches
(https://github.com/FirebirdSQL/firebird/tree/batch) BLOBs have
hard-coded BPB which makes all BLOBs have stream type. Certainly this is
not what we want in release version. I suggest the following:
- current hard-coded BPB remains as default for all BLOBs created by
Batch when other BPB is not set - i.e. with Batch default type of blob
becomes stream one, not segmented;
- function is added to the Batch interface making it possible to preset
any other user-defined BPB for blobs that will not have it set explicitly;
- 2 parameters are added to addBlob() function containg (as always)
length of BPB and it's data, '0, NULL' means use of default BPB;
- BLOB header in addBlobStream() is extended with additional parameter,
BPB length, with BPB data stored right after the header and followed by
BLOB data.
Next we need a way to pass data of segmented BLOB to the Batch. When
using blob stream mode it's quite clear - each segment in the stream is
prefixed by 16-bit length of segment aligned on 2-byte boundary. Length
is provided in host (not network) order, i.e. remote layer should
provide xdr-level conversion. When using addBlob() and appendBlobData()
looks like the only clear for users way to go is to make addBlob() store
first segment with next segments stored by appendBlobData(). That's not
top efficient in case of small segments but luckily BLOBs have buffering
in both engine and network providers. i.e. performance should not fall
seriously.
Comments?
------------------------------------------------------------------------------
Check out the vibrant tech community on one of the world's most
engaging tech sites, Slashdot.org! http://sdm.link/slashdot
Firebird-Devel mailing list, web interface at
https://lists.sourceforge.net/lists/listinfo/firebird-devel
