On Sat, 31 Aug 2019, Benito van der Zander wrote:
Hi,
when I need maximal speed, I use jsonscanner of fpJSON. It should be
vastly faster than a complete parser by not allocating memory or
validating the json.
Although it decodes all the strings in the JSON, which require some
allocations, even if you do not need every string.
It would be faster, if it would just set a (pchar, length) pair for each
string. The CurTokenString property could be changed from returning the
already decoded string to a method decoding the string only when the
property is read.
Indeed, that is also why the JSONParser/JSONReader exists.
It allows a SAX-like approach to JSON, which is really important when dealing
with
huge documents. If you need to insert data at high speed in a database, you
don't need the whole JSON Document which would allocate huge amounts of
memory (at least twice the original JSON stream).
You use JSONSReader and handle record by record.
As I said, the scope of fpJSON is much bigger than the JSONtools approach.
It simply became so out of necessity.
I'm currently implementing a speed improvement, but your suggestion is
interesting and will probably result in even more speed enhancements.
Michael.
_______________________________________________
fpc-pascal maillist - fpc-pascal@lists.freepascal.org
https://lists.freepascal.org/cgi-bin/mailman/listinfo/fpc-pascal
