25303 2014-04-10 01:37:26 +0000 Small bugs in the descriptions 2014-04-17 17:44:00 +0000 1 1 1 Unclassified WebAppsWG DOM Parsing and Serialization unspecified PC Windows NT RESOLVED FIXED P2 normal --- 1 crimsteam travil art.barstow mike www-dom public-webapps-bugzilla oldest_to_newest 103648 0 crimsteam 2014-04-10 01:37:26 +0000 Hi, I gathered together a few minor bugs: ----------- https://dvcs.w3.org/hg/innerhtml/raw-file/tip/index.html#dfn-dom-element-insertadjacenthtml Algorithm for insertAdjacentHTML(position, text) method, step 3: [3. Let fragment be the result of invoking the fragment parsing algorithm with text as markup, and parent as the context element.] We have "parent" but don't know what exactly it is. We jump to other algo (parameter is context element). ----------- https://dvcs.w3.org/hg/innerhtml/raw-file/tip/index.html#h3_parsing In Parsing, first line we have: [The following steps form the fragment parsing algorithm, whose arguments are a markup string and a context element.] Dot (.) at the end should be replaced by a colon (:). And the same for: [The following steps form the fragment serializing algorithm, whose arguments are a Node node and a flag require well-formed.] In the same Parsing algo, but step 5: [5. Append each node in new children to fragment (in order).] and some other place (like "XML serialization algorithm", "produce a DocumentType serialization", "record the namespace information") "in order" should be replaced by "tree order" and linked to DOM spec: http://dom.spec.whatwg.org/#concept-tree-order. This term is used in DOM and HTML5 spec.[1] ----------- DOM in some IDL fragment use extended attributes (like [NewObject] and [SameObject]): [Constructor] interface Document : Node { [NewObject] DocumentFragment createDocumentFragment(); } So, you can do the same for some method: [Constructor] interface DOMParser { [NewObject] Document parseFromString (DOMString str, SupportedType type); }; partial interface Range { [NewObject] DocumentFragment createContextualFragment(DOMString fragment); }; ----------- https://dvcs.w3.org/hg/innerhtml/raw-file/tip/index.html#widl-DOMParser-parseFromString-Document-DOMString-str-SupportedType-type In parseFromString method step 2: [2. If the previous step didn't return an error, return the newly created document and terminate these steps.] Phrase "and terminate these steps" can be omitted because term "return/throw" is unambiguous (leave algorithm). This convention is used in DOM spec. The same in outerHTML step 3: [3. If parent is a Document, throw a NoModificationAllowedError exception and terminate these steps.] and insertAdjacentHTML step 1: [1....If context is null or a document, throw a NoModificationAllowedError and terminate these steps....] So when we return something or throw error writing "terminate these steps" is unnecessary. Link to changelog in DOM about this: https://github.com/whatwg/dom/commit/d621e7e595e05808c6c54ea1cf4930cbfcc6316b ----------- In DOM all error names are written in quotes, so in this spec you can do similarly: NoModificationAllowedError => "NoModificationAllowedError" SyntaxError => "SyntaxError" and all other case. This names may be linked exactly to some place in error names table (http://dom.spec.whatwg.org/#error-names-0) << ale names have id.[1] ----------- https://dvcs.w3.org/hg/innerhtml/raw-file/tip/index.html#h2_extensions-to-the-range-interface In chapter "8 Extensions to the Range interface", we have this code in green box: fragment = range . createContextualFragment(fragment) Will be better if use "var" at the beginning and different name for the variable and argument, now we have "fragment". Just something like this: var documentFragment = range . createContextualFragment(fragment) or without "var" but different name. ----------- Probably I will fine more small things but first I must jump from WHATWG spec ( continuously developed?) to W3C. [1] Of course all suggestions linking to a new DOM can be make for W3C spec version. 103765 1 travil 2014-04-11 16:28:05 +0000 Thanks for the close inspection of the algorithms. I will attempt to clarify these. 104031 2 art.barstow 2014-04-17 12:20:46 +0000 Travis - I'm going to block a CfC for LC until this bug is fixed. Are there any other open issues or areas of contention (that should) block LC? 104036 3 travil 2014-04-17 17:19:14 +0000 (In reply to spiritRKS1910 from comment #0) > Hi, I gathered together a few minor bugs: > > ----------- > https://dvcs.w3.org/hg/innerhtml/raw-file/tip/index.html#dfn-dom-element- > insertadjacenthtml > Algorithm for insertAdjacentHTML(position, text) method, step 3: > > [3. Let fragment be the result of invoking the fragment parsing > algorithm with text as markup, and parent as the context element.] > > We have "parent" but don't know what exactly it is. We jump to other algo > (parameter is context element). I think "parent" is supposed to be "context". Context gets the parent under some of the conditions in step 1. > ----------- > https://dvcs.w3.org/hg/innerhtml/raw-file/tip/index.html#h3_parsing > In Parsing, first line we have: > > [The following steps form the fragment parsing algorithm, whose > arguments are a markup string and a context element.] > > Dot (.) at the end should be replaced by a colon (:). > > And the same for: > [The following steps form the fragment serializing algorithm, whose > arguments are a Node node and a flag require well-formed.] Good catch. > In the same Parsing algo, but step 5: > [5. Append each node in new children to fragment (in order).] > > and some other place (like "XML serialization algorithm", "produce a > DocumentType serialization", "record the namespace information") > > "in order" should be replaced by "tree order" and linked to DOM spec: > http://dom.spec.whatwg.org/#concept-tree-order. This term is used in > DOM and HTML5 spec.[1] Sounds good. I counted 14 occurrences of "in order" but many of these are referring to following a sequence of steps in order. Two require a link to DOM's "attribute list" concept. I'll fix-up the prose to make all occurrences more clear. > ----------- > DOM in some IDL fragment use extended attributes (like [NewObject] and > [SameObject]): > > [Constructor] > interface Document : Node { > [NewObject] DocumentFragment createDocumentFragment(); > } > > So, you can do the same for some method: > > [Constructor] > interface DOMParser { > [NewObject] Document parseFromString (DOMString str, SupportedType type); > }; > > partial interface Range { > [NewObject] DocumentFragment createContextualFragment(DOMString fragment); > }; Super. Will do. > ----------- > https://dvcs.w3.org/hg/innerhtml/raw-file/tip/index.html#widl-DOMParser- > parseFromString-Document-DOMString-str-SupportedType-type > In parseFromString method step 2: > > [2. If the previous step didn't return an error, return the newly > created document and terminate these steps.] > > Phrase "and terminate these steps" can be omitted because term > "return/throw" is unambiguous (leave algorithm). This convention is used in > DOM spec. > > The same in outerHTML step 3: > [3. If parent is a Document, throw a NoModificationAllowedError > exception and terminate these steps.] > > and insertAdjacentHTML step 1: > [1....If context is null or a document, throw a > NoModificationAllowedError and terminate these steps....] > > So when we return something or throw error writing "terminate these > steps" is unnecessary. Link to changelog in DOM about this: > https://github.com/whatwg/dom/commit/d621e7e595e05808c6c54ea1cf4930cbfcc6316b OK. After removing this phrase it only occurs once in the conformance section, and once in an algorithm which does not involve a return/throw condition. Thanks for the pointer. > ----------- > In DOM all error names are written in quotes, so in this spec you can > do similarly: > > NoModificationAllowedError => "NoModificationAllowedError" > SyntaxError => "SyntaxError" > > and all other case. This names may be linked exactly to some place in error > names table (http://dom.spec.whatwg.org/#error-names-0) << ale names have > id.[1] Nice cross-linking. Sounds good. > ----------- > https://dvcs.w3.org/hg/innerhtml/raw-file/tip/index.html#h2_extensions-to- > the-range-interface > In chapter "8 Extensions to the Range interface", we have this code in green > box: > > fragment = range . createContextualFragment(fragment) > > Will be better if use "var" at the beginning and different name for > the variable and argument, now we have "fragment". Just something like > this: > > var documentFragment = range . createContextualFragment(fragment) > > or without "var" but different name. Agree that this explanatory usage isn't as good as it could be. I'll change the name of the variable receiving the return value and the parameter to be less ambiguous. > ----------- > Probably I will fine more small things but first I must jump from WHATWG > spec ( > continuously developed?) to W3C. > > [1] Of course all suggestions linking to a new DOM can be make for W3C spec > version. Thanks 104041 4 travil 2014-04-17 17:42:50 +0000 Fix committed in the latest editor's draft: https://dvcs.w3.org/hg/innerhtml/rev/97f9c81f0990 104042 5 travil 2014-04-17 17:44:00 +0000 (In reply to Art Barstow from comment #2) > Travis - I'm going to block a CfC for LC until this bug is fixed. Are there > any other open issues or areas of contention (that should) block LC? OK, unblocked :-) No other issues I'm aware of.