Print

Print


Tim Spalding wrote:
> Does processing speed of something matter anymore? You'd have to be
> doing a LOT of processing to care, wouldn't you?
>   

Yes,which sometimes you are. Say, when you're indexing 2 or 3 or 10 
million marc records into, say, solr.

Which is faster depends on what language and what libraries you are 
using for both binary marc and marcxml. But in many of our experiences, 
parseing and serializing binary marc _is_ significantly faster than 
parseing and serializing marcxml.  That is of course just one of the 
various criteria that comes into play when choosing a format.

Here's Bill Dueber's benchmarks comparing MarcXML, marc binary, and a 
marc-in-json format; in ruby, using various library alternatives.  I 
rather like the marc-in-json format for being a happy medium.  Whether 
it's "standard" or not doesn't neccesarily matter when you're dealing 
with your own records, passing them through several stops on a 
toolchain, and have tools available that can do it. Who cares if 
any/everyone else uses it.

http://robotlibrarian.billdueber.com/sizespeed-of-various-marc-serializations-using-ruby-marc/