NTriplesParser optimization about rdf4h

[cordawyn]

Sounds great! It's up to @robstewart57 to decide, but I think it would be better if you just created a pull request for your changes - this way we can both see what's changed and discuss stuff in more detail in the comments (and reject your pull request, if it's not good enough 😉 ).

[robstewart57]

Hi @mgmeier I'm very excited to see the work you've put into this, thanks! I'm also snowed under for the next week or so, but after that I'll definitely look in detail at the changes your commits have made, and will provide feedback.

[robstewart57]

@mgmeier interesting approach. First things first. Prior to your commits:

91 out of 781 tests failed (38.00s)

Once you've followed these instructions:
https://github.com/robstewart57/rdf4h#running-tests

How many tests fail?

[robstewart57]

I've merged 76722cc .

This does seem to have triggered some failing tests, 94 rdf W3C tests failed before, now 105 fail. Nevertheless, a 9000% speedup is too good to resist 👍

I have resisted merging the other two commits for now, as it introduces state in the GenParser used for parsing NTriples

I'm going to upload rdf4h-2.0.0 to hackage, which represents a substantially more robust collection of RDF parsers and a slight change in the API in the names of the RDF graph instances.

The important thing now, is to try to squash the 105 out of the 784 failing W3C unit tests.

[cordawyn]

Nice!

On Sat, 16 Apr 2016, 13:14 Rob Stewart, [email protected] wrote:

I've merged 76722cc
76722cc
.

This does seem to have triggered some failing tests, 94 rdf W3C tests
failed before, now 105 fail. Nevertheless, a 9000% speedup is too good to
resist [image: 👍]

I have resisted merging the other two commits for now, as it introduces
state in the GenParser used for parsing NTriples

I'm going to upload rdf4h-2.0.0 to hackage, which represents a
substantially more robust collection of RDF parsers and a slight change in
the API in the names of the RDF graph instances.

The important thing now, is to try to squash the 105 out of the 784
failing W3C unit tests.

—
You are receiving this because you commented.
Reply to this email directly or view it on GitHub
#35 (comment)