1547 2005-07-11 07:11:38 +0000 [FS] editorial: 3.2.1 Processing model 2007-01-16 17:21:21 +0000 1 1 1 Unclassified XPath / XQuery / XSLT Formal Semantics 1.0 Last Call drafts All All CLOSED FIXED P2 minor --- 1 jmdyck simeon public-qt-comments oldest_to_newest 4503 0 jmdyck 2005-07-11 07:11:38 +0000 3.2.1 Processing model "Static analysis is further divided into four sub-phases. Each phase consumes the result of the previous phase and generates output for the next phase. ... Static analysis consists of the following sub-phases 1. Parsing 2. Static Context Processing 3. Normalization 4. Static type analysis" In fact, as section 5 tells us, some Normalization ([]_PrologDecl) happens as part of SCP, and some ([]_sequencetype) as part of STA. Moreover, some STA happens as part of SCP (5.14, 5.15). So "sub-phases" 2, 3, and 4 are not as assembly-line as you indicate. (leftover from last year, comment #029) "Normalization works by bottom-up application of normalization rules over expressions, starting with normalization of literal expressions and variables." I disagree that it's necessarily bottom-up. You could equally say that it starts at the top with an invocation like [[ QueryBody ]]_Expr or [[ PrologDecl ]]_PrologDecl, which then proceeds by invoking successively lower-level rules. "The formal description of evaluation works by bottom-up application of evaluation rules over expressions, starting with evaluation of literals and variables." How can you possibly evaluate a variable reference if you haven't yet evaluated its declaration, which is higher up in the tree? You say "Note that in practice some implementations may prefer top-down evaluation strategies." But if it's a matter of preference, how can you say it's one or the other? "32 bits integers" s/bits/bit/ 4505 1 jmdyck 2005-07-11 19:15:35 +0000 [Re the (non-)pipelining of SCP, Normalization, and STA...] Furthermore, STA is required for some Normalization (4.1.5), so it's even more muddled. 8357 2 mff 2006-02-21 18:53:09 +0000 In 3.2.1, changed text to: "Static analysis is further divided into four sub-phases. Typically, each phase consumes the result of the previous phase and generates output for the next phase. When processing query prologs, these phases may be mutually dependent (See <specref reg="id-query-prolog"/>)." "Normalization works by recursive application of the normalization rules over a given expression." "The formal description of evaluation works by recursive application of the evaluation rules over a given expression." 12184 3 jmdyck 2006-09-29 05:27:42 +0000 (In reply to comment #2) > > "The formal description of evaluation works by recursive application of > the evaluation rules over a given expression." Now that you've removed the reference to "bottom-up application of evaluation rules", it doesn't make any sense to keep the sentence "(Note that in practice some implementations may prefer top-down evaluation strategies.)". Just delete it. 12557 4 simeon 2006-10-19 23:38:09 +0000 Done. - Jerome