124743 - RDF SEQ optimizations

Status: RESOLVED FIXED

(Core Graveyard :: RDF, defect)

Description

[Robert John Churchill]

Two things:

1) of Bag/Seq/Alt, Seq is going to be the the highest frequency of positive
hits... so RDF's container utilities should check for SEQs first

2) code which satisfies all of the following:
     o has a private in-memory datasource
     o uses ONLY RDF_SEQs for containment
     o uses RDFContainerUtils->IsContainer() to check for containment
would benefit from using RDFC->IsSEQ() instead, as it does 1/3rd the work.

Comment 1

[Robert John Churchill]

Attachment: Patch

Comment 2

[Robert John Churchill]

Chase:  care to review?

Comment 3

[Chase Tingley]

Comment on attachment 68797 [details] [diff] [review]
Patch

If you're really trying to squeeze every drop out of IsContainer(), why not
ditch the IsA() calls (each of which calls HasAssertion()) and replace them
with a single call to GetTarget(aResource, kRDF_instanceOf, ...) and  a series
of equality checks?

Other than that, sure.	r=tingley

Comment 4

[Robert John Churchill]

Chase: yeah, I thought of that but was a tad leery as aggregation of multiple
datasources could confuse a single GetTarget() call while multiple distinct
HasAssertion()s might succeed. It is a bit farfetched but again I was leery.

Comment 5

[Chase Tingley]

You're right, there is a bit of an issue with composite datasources.

There'd be more of an issue if we did things correctly and used rdf:type 
instead of rdf:instanceOf to store container type (bug 15006) (since collisions 
between container type and other type information would be more likely -- using 
instanceOf, it should only be possible to collide with *other* container types 
in the case when a resource somehow had more than one.  But getting the correct 
one doesn't really matter for the purposes of determining IsContainer()).  
Either way, GetTargets() should work.  But at that point it may not be faster 
any more (although it might be, going by the immutable law that HasAssertion == 
slow :) ).  I trust your judgement.

Comment 6

[Ben Goodger (use ben at mozilla dot org for email)]

Comment on attachment 68797 [details] [diff] [review]
Patch

sr=[e-mail]

Comment 7

[Robert John Churchill]

Fixed.

Comment 8

[benc]

tever is not RDF QA anymore