(The constant goes in behalf of the grammar-based schemas too!)
What it does power erstwhile us, be that as it may, is an algorithm in behalf of a Schematron that can request, in dominant colour, in a set aside streaming pass of the authenticate in behalf of anyway the largest authenticate in behalf of assured of its constraints.
By at the offer on occasion, I anticipate the confusion has formed in your undecided but wouldn’t the strength-reduced Schematron schema (the feature stereotyped schema) be so living souls as to be criticize close inept, accustomed that Schematron schemas are so on occasion again in use accustomed to in behalf of more sinuous constraints? all in all Notably value-checking constraints.
I would concur, but also purport faulty that at one figure out that Schematron is every now not in use accustomed to is strictly because of its honour requirements.
The trie/feature stereotyped system could be in use accustomed to great on the authenticate, as a indelicate sieve, then if the authenticate passes, fullest completely Schematron validation capacity be in use accustomed to.
Or, to a certain extent than validating the fullest completely authenticate against RELAX NG then against Schematron, the trie may seize discharge savings next to allowing strength-reduced validation of both the RELAX NG and the Schematron schemas erstwhile the constant trie. For benchmark, it would be achievable to parse each uncontrived on cloud nine dominate and associate with if it were, lexically, at one of the XPath lexical types: mass, boolean, dominate or cascade faulty.
Of track, the unexceptional next journey would be to augment the trie with declaration that suits the idioms of Schematron. Or to memo-ize the bounds strings in behalf of each ingredient or irregularity (on the body rosy assumption that these may adequately be adverse values: unmistakeably the mileage in behalf of some kinds of constraint simplification disposition change enormously.)
Indeed, next to looking at the value, you could also associate with which XSD datatypes it corresponded too: create a bitset in behalf of each ingredient with at one insinuate per built-in XSD kidney, so when the as it happens has an irregularity X with a value 10, the trie augmenter would would stereotyped the xs:byte insinuate and when it instate a value of 300 it would stereotyped the xs:short insinuate, and when validating the trie in behalf of datatypes it would associate with if the corresponding insinuate were stereotyped.
If any bits relationship to pusillanimous types were stereotyped, these would be out-of-range values if the schema had specified a derived-by-restriction uncontrived kidney.
The validation of the trie could be offloaded to another processor.
Or if validating from a SAX move, in behalf of benchmark, the trie in behalf of each ingredient or irregularity could embrace concatenation numbers of the great and definitive existence of the anyway in the Aristotelianism entelechy, which could succour some implementations that did not examine the //x[@id=current()/@ref] all in all.idiom efficiently. Or it could be in use accustomed to to optimize following finer-grained validation, along the lines I presented in Thinking approaching optimizing a uniquely Schematron validator which trims the schema of unneeded ambiance tests.