[NASPRO-news] I can read you, Turtle!
- From: zanga.mail@xxxxxxxxx
- To: naspro-news@xxxxxxxxxxxxx
- Date: Wed, 3 Nov 2010 01:07:11 -0500 (CDT)
This has been a less intensive coding week than the previous, yet a major goal
was reached: the NASPRO Turtle parser is completed and is fully conformant,
according to the test suite distributed with the Turtle - Terse RDF Triple
Language specification.
Yet, its performance is quite poor as of now. This is not due to the parser
code in itself, which instead is capable of very high throughput, but to the
underlying graph implementation in NASPRO core. In particular the bottleneck
resides in the use of simple doubly-linked sorted lists holding nodes and
edges. Hence, the plan is to replace them with AVL trees as soon as possible,
whose coding is already in a good stage.
While developing the parser, a number of ambiguities and errors were found in
the original Turtle grammar specification, all of which, together with the
relative solutions adopted in our parser, have been reported to the authors.
In the end, I am having an interesting discussion on the LV2 development
mailing list with Dave Robillard, author of LV2, SLV2 and Ingen, to name a few,
regarding the implementation of data retrieval, querying and storage for
supporting LV2 plugins. Whatever will come out of this, I want to thank him for
being open minded enough to see my effort from the constructive, rather than
competitive, point of view.
Other related posts:
- » [NASPRO-news] I can read you, Turtle! - zanga . mail
