CVS log for 2006/webapi/XMLHttpRequest/Overview.src.html

Up to [Public] / 2006 / webapi / XMLHttpRequest

Request diff between arbitrary revisions

Keyword substitution: kv
Default branch: MAIN

license

define header manipulation in terms of author request headers to work better with CORS

update method and constructor terminology (more terse)

each XMLHttpRequest object needs its own task source

throw for synchronous usage of new features in Document (i.e. non-Worker) context

convert asynchronous flag into an synchronous flag for an upcoming change

allow withCredentials to be set during UNSENT

more clearly distinguish between simple state changes (change) and state change steps (switch)

ack Ms2ger

align event handler attributes with the new ugly-syntax reality http://www.w3.org/Bugs/Public/show_bug.cgi?id=14351

reference exceptions

align with DOM4

event listeners do not have to be on both objects for it not to be garbage collected

url is not optional; doh

garbage collection did not account for XMLHttpRequestUpload

use optional rather than overloading

do not not mention XHR2 events in XHR1

attempt at defining garbage collection

use concept-document

consistify

ED

more cleanup (Anolis)

remove IDs as Anolis generates them

XHR uses Anolis now

RFC can haz cookies

restrict overrideMimeType in the same way responseType is restricted

set the boundary string for multipart/form-data; thanks Hixie

define Window/Document by reference

reign of fire

make Terminology a top-level section and rename Conformance Criteria to Conformance for consistency

exceptions are defined by DOM Core now

Window Object draft needs no mention; we need these definitions beyond HTML5

there is no mediaType; only dispatch progress events for async calls

define the origin of Document objects returned by responseXML

cookie-free now comes for free so no need to say it

arraybuffer is here to stay, is what the people say

define responseType/response

remove followRedirects for now

remove leading U+FEFF character; doh

abort() can be terminated too

let the various response* attributes also use the error flag; allows removing other things; no functional change

more minor editorial nits

fix minor issue with authentication text and simplify it as well

address various editorial comments

do something before terminating

consistify spacing of examples

XHR2 will not have exit criteria for now

introduce followRedirects; there will be much rejoicing

introduce responseBlob member

request timeout and credentials flag are to be set during open()

put things in logical order

set media type correctly for HTML documents; fixes http://www.w3.org/Bugs/Public/show_bug.cgi?id=10324

special casing URL schemes is for noobs; fixes http://www.w3.org/Bugs/Public/show_bug.cgi?id=10327

converge same-origin rules between XHR1 and XHR2; risky fix; fixes http://www.w3.org/Bugs/Public/show_bug.cgi?id=10322

everyone does XML; fixes http://www.w3.org/Bugs/Public/show_bug.cgi?id=10325

no longer mention Accept-Encoding and Accept-Charset; fixes http://www.w3.org/Bugs/Public/show_bug.cgi?id=10328

Set-Cookie2 is not part of the web; no need to talk about it in the send() section

clarify a state transition by making things more explicit

fix typo

fix internal references for inflating and deflating

getAllResponseHeaders() needs to link too

user credentials is only used in XHR2

sync source doc with CR publication

minor CR fixes

whee CR

forgot a link in previous versions

add link to disposition of comments

point to test suite; point out minor issue with responseXML

sync wording with CORS

copypasta

make a change Art proposed to #crec

add security document as CR exit criteria

cosmetics

remove two bogus sentences; add a bunch more intro; improve wording of a note a little

more cosmetic changes; this time setRequestHeader

more domintro magic; also nuking some old stuff

domintro readyState

domintro for the status and response attributes

domintro abort()

editorial open method

and more editorial stuff; now events

fix the states

introduce title= for send()

change IDs now we still can

domintro trial

add candidate recommendation exit criteria

remove dates, editions, and the term editor frtor from references

AnonXMLHttpRequest()

no cookies for you, Document

oops; add boolean in front of constructor argument

new XMLHttpRequest(anon)

fix authorization stuff a little better

separate Fetch and request URL with a word

start using fetch

turn evil method should into must; turn banned header should into must; fix some minor issues

reword definitions per bz

fix a few things related to redirects

move from Unicode to bytes; fix Olli\'s name

update response entity body using the event queue

actually, lets not; also actually terminate the request in case of network errors

lets make it explicit

empty the task queue

only need one task source after all

clean up a bunch of things; also unicorns

remove a bunch of nested 'queue a task'; still need to solve race condition and updating of the response entity body

generate XHR1 and XHR2 from a single file

markup typo; add Huub

fix type; back to ED

copy pasta

HTML4 is dead to me, but not to almighty pubrules I suspect

prepare for publication; remove outdated comment

typo

merge near identical algorithms into one

tiny editorial changes

Implement resolution of http://www.w3.org/2008/webapps/track/issues/103 and remove unneeded markup; also remove one check that is no longer needed in getResponseHeader()

nit

put Document before DOMString

add reminder so I can remove emails

mention domintro boxes (will be added during CR or so); remove obsolete advice

error handling in case of multiple charset parameters

DOMString != Unicode

Content-Type is not changed when there is no charset parameter; also rewrite how I dealt with this; DOM Events is no longer namespaced

improve requirements on Accept and Accept-Language

make the XHR specs pub ready

fix copyright date and titles

WebIDL cleanup

use end user for clarity

use AllowAny

defining this properly is simpler in terms of dom-innerHTML

fix a small error

editorial consistency with other specs

define status and statusText like every other member

cleanup

simplify send() algorithm some more

use the same terminology as in XHR2

Allow the Authorization header to be set; it is safe cross-origin

minor editorial changes

split out cookies and switching to the done state

oops

fix small mistake

make the writing match XHR2 closer to make it easier to keep the specs in sync

put request event rules into its own section

restructure the async/sync part of send()

method token is a production

define simple terms inline

do not rely on innerHTML but rather the XML fragment serialization algorithm

fix sotd; simplify title

markup typo; introduce Function from HTML5

link from IDL to event handler attributes, states, request, and response; also put sections in order

group all request variables; also introduce request entity body

define base URL and origin better; integrate with the event loop and storage mutex; add more sections as some people requested; rewrite part of the send() algorithm because of the event loop; sorry about the diff

remove silly use of quotes; clear error flag when invoking send()

hide Set-Cookie and Set-Cookie2 from response headers

reference HTML5 for URL, resolves issue with URLs not following IRI syntax; also reference HTML5 for case issues, simplifies things; make it clear when user agents must not prompt the user based on IRC conversation with sicking; update Web IDL reference

change exception codes to be more consistent

add several pointers to US-CERT for the disallowed  methods

only require the basics of the Window object...

add comma, add fails, add ed

clarify a note; add another reference to XML

send() with and Content-Type changes; now also for DOMString and also charset fixup for both Document and DOMString

remove confusing paragraph in events section; add a note about exceptions

make getAllResponseHeaders a little bit more clear; mention the character name after the Unicode reference; make a small editorial change to the note at the start of send()

apply IRI -> URI conversion

remove encoding of user/password; leave it up to other specs

add link to latest editor version; make several editorial fixes

HTML5 dropped UTF-32; makes sense

reference the IRI thingie

make it more clear that failure does not result in user interaction

make handling 401 responses match IE; address comments from MC

use HTML5 same-origin; define author request headers; don't allow UAs to set headers other than those explicitly permitted by the spec (should-level though)

treat GET/HEAd the same; drop entity body for both

editorial changes to the dependencies section; also reference the correction section in HTML5

integrate Web IDL; serialize using document.innerHTML

make requirement for Accept:*/*

fix several examples; clarify some text

add 2109 back even though cookie specs are useless

add a reference to HTTPONLY cookies

add a note for TRACK; add Authorization / Cookie / Cookie2; add TLS negotiation failure; rephrase DOM requirements

make several small editorial changes per LC comments

make Accept always present; explain error/send() flags

fix two typos; make ED draft

prepare for Last Call publication

return null for invalid headers in getResponseHeader()

getAllResponseHeaders() returns the empty strng in case of failure/non same-origin

update the XMLHttpRequest stuff to not throw exceptions for responseText and responseXML; responseText is always a DOMString too

ban headers starting with Sec-

responseText shall return the empty string

GET will not have an entity body

use inputEncoding for encoding a Document

ensure charset is set to UTF-8

make 'conforming xml user agent' more 'clear' plus several non-normative changes

set accept-language conditionally

accept-language is not to be set by the UA

fix DOM2Events ref

do acknowledgments different

HttpOnly; remove duplicate stuff

use event handler attribute dfn from HTML5; document pointer

make the method story with respect to security more clear

DOM3Events -> DOM2Events; clearly mark dependency on HTML5; remove dependency on obsolete Window spec

reword

change case-insensitive dfn; add security consideration sec

no need for a base style sheet

make draft ready for publication; remove note I missed earlier

make some editorial changes

header not match syntax production -> empty string

typo

remove data: URL support for now

remove obsolete reference

solve last issue with respect to same-origin and make some editorial changes

define which Document object same-origin is checked against

use term ihost; can't use baseURI; clearly mark open issue about which Document object to use

reference HTML5

define same-origin

define XML user agent better; small editorial change

less UTF-8; note work in progress in references

clarify how exceptions work; introduce SECURITY_ERR definition into this draft to guide implementors

fix typos

The anti-UTF-8 dev.w3.org movement has won.

set send() flag to 'false' later on in the abort() algorithm to ensure the algorithm functions properly

s/OPEN state/OPENED state/ to avoid case issue with IDL; minor editorial fixes

add latest 'previous version'

disallow anything that starts with 'Proxy-'

improve fragment identifiers

forgot EH; editorial changes

block Proxy-Connection header for security reasons

add a note saying that responseXML just gives you a (static) tree

make it an editor's draft again and make several editorial changes

make several editorial fixes at request of Denis

add Stewart to the list of acknowledgements

introduce error flag to more clearly indicate behavior; fix getResponseHeader() and getAllResponseHeaders()

introduce HEADERS_RECEIVED

fix CSS and update date as requested by W3C webmaster

introduce ABORT_ERR; hopefully last new feature needed...

make it a W3C WD ready for publication

introduce class=switch

high potential for errors: make the send() algorithm more explicit and add text/html charset detection

fix some typos; fix SENT; remove some incorrect EOL

drop text/xsl

make a small clarification to the charset detection algorithm

mention overrideMimeType

simplify charset detection for text response entity body

forget a bunch of people in the list of acknowledgements

cover BOM detection in text response entity body

s/stop/terminate/

fix markup error

clarify how the send() algorithm might get aborted and make it a bit more explicit

change some of the cross references around response entity body and remove a redundant paragraph

410 Conforming script

markup fix

make it more explicit when the send() flag is set to "false"

make XML response entity body steps slightly more clear and fix the send() flag pointers

elaborate on non same-origin; change when the send() flag is set; fix an obvious error with the default value of async

throw NOT_SUPPORTED_ERR for unsupported schemes, fix two (editorial) mistakes

grr

address comments from Bjoern; don't dispatch event after an exception and allow responseXML for no Content-Type responses

terminate the phrase 'abort these steps' in favor of 'terminate these steps'

doesn't -> does not; don't -> do not

remove lots of redundant conformance requirements; thanks Simon for pointing that out!

clean up 'Conformance' a bit more

introduce Conforming XML user agent based on a comment from Maciej

make some more minor comments to satisfy a comment

make some editorial changes based on feedback from Mohamed on public-webapi

add text/xsl and fix an editorial issue; also update DoC

remove some old stuff and define charset sniffing for responseText

reset response entity body as opposed to the attributes

define response entity body, text response entity body and xml response entity body

define send() flag up front

change RECEIVING to the simpler LOADING

when the state is DONE don't dispatch readystatechange

editorial: simplify some language

when abort() is invoked you don't always switch to DONE and dispatch readystatechange

fix editorial typos and make sure state switching happens synchronously as well during send()

make the last two steps of responseXML more clear

define case-insensitive match

s/LOADED/DONE/ as that better reflects the state

more formatting

editorial

formatting fix

first step at fixing responseXML; mark up a MAY

fix some conformance requirements based on a comment from Simon Pieters; add a note about addressing case-insensitive

typo; thanks bjoern

fix some typos pointed out in LC comments; fix readystatechange, statusText and status based on comments from Hixie

several typos + s/UNINITIALIZED/UNSENT/

change the layout of the "IDL", fix an issue with setRequestHeader (value or header being null)

No need to list Date twice + editorial

add four more headers that need to be ignored for security reasons and add a list of other HTTP methods that potentially need to be uppercased

make sure readystatechange is dispatched after send() is invoked as well and fix abort()

editorial

improve variable names and say that people should support XML for interop reasons

some more editorial; also fix status to say editor's draft

some more editorial; also fix status to say editor's draft

fix the wording of the readystatechange event

editorial fix

fix abort()

change setRequestHeader() to do what IE does

handle method invocation as a set of steps to ensure correct order of dispatching of exceptions and to make it more clear what is expected; be more clear about when events are to be dispatched; introduce constants; merge section 2.0 and 2.1; my apologies for doing this all in one go; more changes to come though

lots of small editorial changes

change the ping to a log sample to not confuse people

fix some more comments from Dan Winship

fix several reported issues

address an editorial remark and add a disposition of comments file

it's not informative...

s/abbr/acronym and change the date manually in Overview.html to make it ready for publication

remove some bogus statements made in responseText; should prolly make more testcases on this; should also reply to bjoern

remove reference to an obsolete cookie draft; we already referenced a newer version

address some comments from bjoern: don't talk about things already specified somewhere else

pleasing pubrules

attempt to make the sentence more readable

solve the user and password encoding problem, for now; say what setRequestHeader does; make sure UAs can either use multiple headers, combine the values, or a combination; use namespace well-formed throughout the draft

move the note to directly after the list of headers and fix the text a bit (thanks bjoern!); also make another small editorial change

add a note to make it clear that people can't reset the values of certain headers

add OPTIONS, move references to the end of paragraphs

implementations may support XML; reference the XML specs

address some more editorial bits, clarify responseText and responseXML and add two more feature requests

address some editorial nits

improve the user and password story

change the way references are done; add dom core as a dependency

small sotd changes

remove some stuff about the spec not being stable, still need to update sotd a bit more; changed wording in a sentence to satisfy comment from Julian and Maciej

fix a small error with null versus empty string

make some small editorial changes

make the list of headers actual list for easy scanning and remove 'Expect' from the second list since it was already in the first

introduce XMLHttpRequestException.NETWORK_ERR

implement a solution for user/password -> last red blob gone; delinkify some stuff; time for more implementation experience!

add dependencies, nuke history as it's not very good and move 'not in this specification' to the end as an appendix

delinkify; point to the [Window] spec

resolve password issue; resolve async network error issue, reformat some things

fixed open() to do what implementations should actually implement

try to implement some alternate solution for the uri resolving mess called 'Window pointer' and rewrite redo the object states

fix the fragment identifier problem, put in a potential solution for open() and make some editorial changes

remove some old stuff

switch to the new postprocessor

define base URI more properly and move some sections around, add a new issue and remove a bunch of methods that really shouldn't be in here

implement a fix for ISSUE-91 and ISSUE-92 and use some suggested text from mjs for resolving URIs

adding more HTTP methods and references

type caused linking not to work

fix an issue and more clearly indicate others

moving the spec style sheet inline and removing some whitespace

fix the way send() works and add the great Alexey!

add some examples, clean up some references and try to implement various resolutions

implement more resolutions and fix the examples

removing the security section and addressing some of the resolutions; more to come...

cleaning up references, still needs lots of work, but checking this in anyway

more editorial work, exposing some issues and such

fix some stupid editorial bugs

add mousewheel.txt

several attributes are really readonly, just like in the WHATWG draft and as sort of implemented...

make it an editor's draft instead of a w3c working draft

remove useless distintion between methods and attributes and put them in IDL order (which also happens to make a lot of sense)

some markup nits

make the specification more readable step 1

actually link to that style sheet...

add source file and some CSS