Posted: Thu May 12, 2016 11:53 am Post subject: On demand parsing BLOB/RCD
Novice
Joined: 22 May 2014 Posts: 14
I think I may be missing a subtlety here...
I'm observing an unexpected scenario with parsing. I have a flow (FileInputNode -> Compute -> MQOutput) within which I am processing a 200MB+ file with Records and Elements set to whole file.
I'm using the large messaging technique in the compute node which is working very well.
If I set the domain to DFDL on the FIN, memory usage only goes up by a miniscule amount (~60MB). However, if I set the domain to BLOB on the FIN and then use an RCD before the Compute node, memory usage goes up by about 700MB. Is this simply to be expected?
Thanks for your response. That's what appears to be happening. However, I thought that RCD does not actually instigate an immediate parse. The code in the compute module has not changed so should still be on-demand...
A streaming parser does not load the entire input bitstream into memory before starting to parse. It loads small chunks, and when each chunk is exhausted it discards it and loads another.
BLOB is not a streaming parser. DFDL is. Does that explain it?
You cannot post new topics in this forum You cannot reply to topics in this forum You cannot edit your posts in this forum You cannot delete your posts in this forum You cannot vote in polls in this forum