All code text in comments should be enclosed in {@code ...}.
Move such extensions to a separate extensions.js.
Really, to follow ES5, Date.prototype.toISOString
We may deprecate this in favor of <pre>
Analyze the security implications
Detect whether this is a valid constructor
walk prototype chain.
Object(<primitive>) wrappers should also be
Having a fastpath flag for this probably doesn't
We should probably
Given that we maintain the invariant that
Given that we maintain the invariant that
Should add a debugging flag to enable
Should
To avoid accidents, <tt>markXo4a</tt>,
non-user-hostile error message
Once ES5 settles, revisit this and
Once we have ES5ish attribute control, it
Revisit this case
Find out if this is still the plan.
consider
domita would currently generate this warning,
See above TODO note.
The only remaining use as of this writing is
Not clear that this is the performant
revisit once we do ES5ish attribute control.
consider returning a defensive copy rather than
Add case for dotted path expression.
encode (or at least sanitize) key
to be read in from a JSON config
If we ever turn on taint checking for
Need a general way to expand readModifyWrite lValues.
These rules should be independent of whether we're writing
Permit should generate a JSON AST directly,
Need a general way to expand lValues
Need a general way to expand lValues
Need a general way to expand readModifyWrite lValues.
Need a general way to expand lValues
Need a general way to expand lValues
Need a general way to expand readModifyWrite lValues.
Nothing is gained by having this be a partial override of
Come up with some notion of a keeper chain so we can
We should probably
Given that we maintain the invariant that
Given that we maintain the invariant that
We may deprecate this in favor of <pre>
Not clear that this is the performant
The only remaining use as of this writing is
Find out if this is still the plan.
Once EcmaScript and Valija allow patterns in parameter
If the resolution of bug #814 is for
figure out why things break when the
Investigate why things break if the following line
Why are we testing for the empty string here?
fix this once DONTENUM properties are better
failure should no longer be an option on (S)ES5/3.
failure should no longer be an option on (S)ES5/3.
failure should no longer be an option on (S)ES5/3.
failure should no longer be an option on (S)ES5/3.
This test isn't green because we haven't yet fixed the bug.
Need more tests.
Fix when bug 956 is fixed.
Fix when bug 953 is fixed.
Fix when bug 953 is fixed.
Fix these tests to test for the output we now expect.
Fix when bug 953 is fixed.
Need more tests.
Remove the concept of "synthetic" entirely.
also need to test when onload isn't set
Would a {@link com.google.caja.reporting.MessageType} be
Fix the "unrecoverable" error responses throughout
Missing tests in CajitaTest.java for the functionality
importedVariables is only used by the root-most scope; it
Uses private access to 's' which is of same class but distinct
Uses private access to 's' which is of same class but distinct
Refactor to use standard Caja Visitor. Currently not
Change the ParseTreeNode type for the right hand sides
Add metadata for a rule indicating that it always fails,
originalSource
sourceLocationMap
imports
manifest
As a special case, an Identifier with a null value is
originalSource
sourceLocationMap
imports
manifest
This class relies on
We do *not* want to support multiple HTML files
http://code.google.com/p/google-caja/issues/detail?id=994
http://code.google.com/p/google-caja/issues/detail?id=994
Will add UncajoledModule wrapper when we no longer
Investigate why this is the right criterion to use.
Our implementation of getAttribute (and friends)
Come up with a uniform convention (and helper functions,
Improve implementation (interleaving, memory leaks)
Do we need more attributes of the event than 'target'?
send()-ing an empty string because send() with no
Expect tamed XML document; unwrap and send
Implement a taming layer for XML. Requires generalizing
Does this work on IE, where console output
Due to the following bug:
Implement
Build a more sophisticated virtual class hierarchy by
Need to pass in the URI rewriter from the build
How do we emulate 'useCapture' here?
How do we emulate 'useCapture' here?
In un-cajoled code, use ___ instead of _ to indicate
Support sending more (variable) arguments with fire()
makeProvider: makeProvider, */
Enhance test framework to allow "before" and "after"
The below should throw MessageType.MASKING_SYMBOL at
Once permit templates can be loaded dynamically, create
The below test is not as complete as it should be
SECURITY: Re-enable by reading (say) x.foo, and
SECURITY: Re-enable by reading (say) x.foo, and
Refactor tests to use checkAddsMessage(...) instead
Change dependents to use checkAddsMessage and
Change checkAddsMessage and similar functions to check
Add negative tests when virtual hierarchy is improved:
Change tests to use structural equality (via quasi
Change CajaArguments to handle >1 occurrence of arg
Use a standard class or move to appropriate package
The service like the gwt version should accumulate
Clarify the meaning of (a||b)* and modify this function
URIs in some contexts (such as links to new pages) should
Outline this and all the other examples of cajole
Fetching ought to be done via a separate service
rewrite attributes
Consider inlining into Playground.html to eliminate roundtrip
Find a nice collection of "typical" html files!
Should whitespace after the units in quantities be escaped
These tests were generated by the fuzzer - distill them to
explain static (sync) loading module id semantics.
justify this set of refs as being sufficient to do
the above probably does the wrong thing in the case where
allow subbing module id resolver
Using XHR loader didn't work; why?
This needs to be redefined in terms of effect/loader;
do we want to reject multiple attach attempts?
At least, that used to be true. The info on that page is tied
have a sensible default instead of this app needing it
Use URL Fetch async requests for parallelism/network
complain to app engine about not being able to
when not debugging, DO put in cache with a shorter timeout
do this with nice output without using jsUnitCore internals?
When the whole URI-policy-implementation thing is cleaned
Figure out why RhinoTestBed.runJsUnittestFromHtml isn't
deploy the already-configured war instead of manually
make this a cajita-uncatchable exception
make this a cajita-uncatchable exception
make this a cajita-uncatchable exception
make this a cajita-uncatchable exception
Figure out a way to leave key alone and
Use fastpath
Use fastpath.
Use fastpath.
Do a lighter-weight wrapping when the function
Only use closure if it's really recursive.
Add code to list all the other handled stuff we know
Add code to list all the other handled stuff we know
Add code to list all the other handled stuff we know
Deprecate this API, since it requires that we leave
Move as many of these as possible to
This test should be replaced by ones for the proper
how to collect logging.
should userException be defined as identity in
Once IE allows escapes inside URLs, then
stop using java.net.URI and use a decent URL
rewrite with a transition table or just use ANTLR
CSS2.1 has changed lexical conventions to effectively
all clients should pass in a proper queue
rename this to CommonMessageType and rename
this is not a valid assumption.
replace "may be null" and "not null" shorthands with
replace with com.google.common.Predicate and the
divert IllegalArgumentExceptions to MessageQueue
divert IllegalArgumentExceptions to MessageQueue
divert IllegalArgumentExceptions to MessageQueue
Is there any value in enumerating elements?
skip if the self name is never used.
can this move from the superclass into a sibling class?
check after Kona meeting
better lower limit
better lower bound
Figure out when post increments are being
How does this work? Why not return node?
once decls consolidated, no need to add to
limit
limit further
Limit further
better lower limit
better lower bound
Figure out when post increments are being
How does this work? Why not return node?
once decls consolidated, no need to add to
limit
limit further
better lower limit
better lower bound
Figure out when post increments are being
check against XML&HTML definitions of whitespace.
maybe enforce the convention that there are matched
breaksReaching should probably take a parameter of type
rename breaks and continues to collectBreakTargets and
Investigate whether use of continue to a non loop
Do we want to remove labelling of non-loop statements
Erase the distinction between LabeledStmtWrapper and
is this true?
this should never be rendered since rendering a group
maybe reliably throw an exception type, that includes
this could probably be more simply done as a
build up a list of classes and ids for use in
this is problematic for DOM nodes parsed without
do we need to emit a static attribute when the
can we get rid of this noop by getting rid of the
make this subclassable so opensocial specific flags can be
remove once JS uri policy is okay
validate sanitizer regexs against the HTML5 grammar at
relying on sanitizeAttributes not to
integrate cajita compiler to allow arbitrary
determine mime type properly.
make sure it really is a DOM node
upward navigation breaks capability discipline.
make nodeClasses work.
make sure event doesn't propagate to dispatched
make sure event doesn't propagate to dispatched
create a proper class for BODY, HEAD, and HTML along
remove these, and only expose them via window once
figure out a mechanism by which the container can
Implement in terms of
define on prototype.
check argument if child now a FunctionCall
for content: and other URI types, use
limit to JSON
should we whitelist nodes here, to e.g. prevent
can bubbling and cancelable on events be simulated
when we change event dispatching to happen
this is currently 6 instead of 4 because it also limits
don't need a list here
the spec is not clear on this. Treat the week
filter byWeekNo and write unit tests
filter byWeekNo and write unit tests
if count is large, we might try predicting the end
warn
if we allow iteration more frequently than daily
apply byhour, byminute, bysecond rules here
look at tzid
translate to start tzid
handle multiple RRULES, EXRULES, RDATES, EXDATES
if there's an easy way to interleave bits when
resolve the href attrib relative to globalProps.url?
resolve the href attrib relative to globalProps.url?
other url node/attrib pairs?
write me
write me
according to section 4.3.10
this may be premature. If needed, We could
extract this a better way once we're providing module
20 is a tuning parameter.
what do we do about the fact that (0 === -0)?
Is this useful? When, in a comparison,
use the message queue passed to the ctor once Scope no
if c is simple and not a global reference, optimize
limit to local references not in with blocks.
eliminate dead branches in if statements.
simplify empty loops
fold string appends in + statements
choose a guaranteed non-interfering name.
refactor this
check locals and formals used
replace with jsdoc comment parser
SVN has it in svn:mime-type, but that is not
should work if there is a break here.
better file positions for error messages.
cajole the scripts so that for loops work. -->
cajole the scripts so that for loops work. -->
cajole the scripts so that for loops work. -->
cajole the scripts so that for loops work. -->
cajole the scripts so that for loops work. -->
cajole the scripts so that for loops work. -->
check that these answers are right
test getSymbol
test rendering of @imports, @page, @font-face,
test rendering of unicode range literals.
this error message seems to be a bug.
better comment each of the test input files.
move parameters into a separate namespace so they
fix message. "fixed" is well-formed but disallowed.
We could break the position rule into multiple
rewrite XHR and setTimeout tests to use this scheme.
write this properly.
enable for all node-types once EQ fixed on IE
enable for all node-types once EQ fixed on IE
This is wrong because we don't allow the insertion point
test that getElementsByTagName and getElementsByClassName
maybe actually diff using
maybe actually diff using
maybe replace this with the JSR 223 stuff.
move unescaping to Escaping.java -- nobody will look there
implement hourly iteration
implement minutely iteration
implement minutely iteration
is this right?
check advancement of more examples
Once they decide on scoping & initialization rules, test
validate the response before eval it
can this first clause go away?
If we ever have a cheap way to test whether result is used freely
move this into scope if we use a single CajitaRewriter to rewrite
support getters and setters in object quasis
move this into scope if we use a single ES53Rewriter to rewrite
(metaweta) Implement block scoping.
(metaweta) Implement immutability auditor and then only allow
It's not masking, it's replacing; invent a different error.
use a whitelist, deactivate the rest by replacing with
do away with the below once Shindig has done away with Neko.
do away with these once shindig has gotten rid of Neko
audit all places where (instanceof Expression) is used since we have
instead of using a block, introduce a statement collection as a super
Once I have network access, figure out if this is here because of a mis-merge?
output to a message queue
π is different from Π
We could append the cajoled HTML to existing contents of the
If they are not under the same virtual doc root, return
If they are not under the same virtual doc root, return
maybe whitelist the color names defined for CSS if the arg is a
expose a read-only version of the document
need a testcase for this
Put a class on this so if the host page cares about 'all iframes'
Safari 2's defaultView wasn't a window object :(
propagate error message
review error handling -- are errors propagated and descriptive?
does not support dependencies. Needs to tie in to
INSECURE TODO INSECURE TODO INSECURE TODO INSECURE TODO INSECURE TODO INSECURE TODO INSECURE
INSECURE TODO INSECURE TODO INSECURE TODO INSECURE TODO INSECURE TODO INSECURE TODO INSECURE
INSECURE TODO INSECURE TODO INSECURE TODO INSECURE TODO INSECURE TODO INSECURE TODO INSECURE
INSECURE TODO INSECURE TODO INSECURE TODO INSECURE TODO INSECURE TODO INSECURE TODO INSECURE
This implementation does not correctly handle [:Cf:] characters.
check for apparent end tag
escape format control characters
unsafe, need to check mimeType but that's not sufficient
the things that need to be TODOne.
|CAVEAT)\b/g);
' || part === 'CAVEAT') {
pull out error handling
change flatten to start anew at every node
finish
implement
implement
add check to make sure that target is a function
implement
implement
implement proper Levenshtein edit-distance
parse @type and @this annotations even without {}s
handle the first declaration in a multi-declaration
handle getters and setters
we can safely inline the reads that are okay.
check that reads are in the same function
don't do this.
use a JSON only lexer.
parameterize
untameRecord isn't defined yet
Can all this JSON stuff be moved out of the TCB?
Should nonextensible objects be stampable?
The latest draft errata fixes this. We'll be ratifying
Where this is used, do we really want frozenness
the positions below seem to be wrong
make an ASCII-art Rhino.
Could optimize out foo.
- do we need a special test for this? is x2v working"
- wait on MarkM to figure out what\'s up with this test."
not sure about this test
- not sure about this one -->",
- I don\'t think attendee stuff is implemented in X2V"
- should \'REQ-PARTICIPANT be lc\'ed? -->",
- should mailto: be here? -->",
- ask MarkM about this. Did he mean to use <del>? -->",
- review the x2v formatting stuff for DESCRIPTION"
-->",
- eh? -->",
*/
*/
*/
Why is this here?
should be weakly keyed, but only used for short lived tests.
Refactor the test cases so that we can use CajitaModuleRewriter
async requirements are not counted in the test status.
webdriver doesn't have a good way of testing relatedTarget.
this test often causes ie6/ie7 to crash. figure out why
can't verify focus changes in webdriver, because webdriver runs
create jsunitRegisterAsync(), then make passes implicit.
print a warning here?
arrange to reliably detect loading errors in loadCaja and report
implement
implement
add check to make sure that target is a function
implement
implement
implement proper Levenshtein edit-distance
parse @type and @this annotations even without {}s
handle the first declaration in a multi-declaration
handle getters and setters
pull out error handling
finish
change flatten to start anew at every node
refactor this
check locals and formals used
replace with jsdoc comment parser
we can safely inline the reads that are okay.
check that reads are in the same function
20 is a tuning parameter.
choose a guaranteed non-interfering name.
use the message queue passed to the ctor once Scope no
what do we do about the fact that (0 === -0)?
Is this useful? When, in a comparison,
if c is simple and not a global reference, optimize
limit to local references not in with blocks.
eliminate dead branches in if statements.
simplify empty loops
fold string appends in + statements
use a JSON only lexer.
parameterize
don't do this.
SVN has it in svn:mime-type, but that is not
how to collect logging.
should userException be defined as identity in
explain static (sync) loading module id semantics.
validate the response before eval it
All code text in comments should be enclosed in {@code ...}.
Missing tests in CajitaTest.java for the functionality
Move such extensions to a separate extensions.js.
Really, to follow ES5, Date.prototype.toISOString
We may deprecate this in favor of <pre>
Analyze the security implications
Detect whether this is a valid constructor
walk prototype chain.
Object(<primitive>) wrappers should also be
Having a fastpath flag for this probably doesn't
We should probably
Given that we maintain the invariant that
Given that we maintain the invariant that
Should add a debugging flag to enable
Should
To avoid accidents, <tt>markXo4a</tt>,
make this a cajita-uncatchable exception
make this a cajita-uncatchable exception
non-user-hostile error message
make this a cajita-uncatchable exception
make this a cajita-uncatchable exception
Once ES5 settles, revisit this and
Once we have ES5ish attribute control, it
Revisit this case
Find out if this is still the plan.
consider
domita would currently generate this warning,
See above TODO note.
The only remaining use as of this writing is
Not clear that this is the performant
revisit once we do ES5ish attribute control.
look at tzid
translate to start tzid
handle multiple RRULES, EXRULES, RDATES, EXDATES
according to section 4.3.10
resolve the href attrib relative to globalProps.url?
resolve the href attrib relative to globalProps.url?
other url node/attrib pairs?
check for apparent end tag
escape format control characters
if there's an easy way to interleave bits when
this may be premature. If needed, We could
write me
write me
don't need a list here
the spec is not clear on this. Treat the week
filter byWeekNo and write unit tests
filter byWeekNo and write unit tests
if count is large, we might try predicting the end
warn
if we allow iteration more frequently than daily
apply byhour, byminute, bysecond rules here
This implementation does not correctly handle [:Cf:] characters.
At least, that used to be true. The info on that page is tied
Use URL Fetch async requests for parallelism/network
complain to app engine about not being able to
when not debugging, DO put in cache with a shorter timeout
have a sensible default instead of this app needing it
unsafe, need to check mimeType but that's not sufficient
In un-cajoled code, use ___ instead of _ to indicate
Support sending more (variable) arguments with fire()
makeProvider: makeProvider, */
INSECURE TODO INSECURE TODO INSECURE TODO INSECURE TODO INSECURE TODO INSECURE TODO INSECURE
INSECURE TODO INSECURE TODO INSECURE TODO INSECURE TODO INSECURE TODO INSECURE TODO INSECURE
INSECURE TODO INSECURE TODO INSECURE TODO INSECURE TODO INSECURE TODO INSECURE TODO INSECURE
INSECURE TODO INSECURE TODO INSECURE TODO INSECURE TODO INSECURE TODO INSECURE TODO INSECURE
Consider inlining into Playground.html to eliminate roundtrip
Fetching ought to be done via a separate service
rewrite attributes
extract this a better way once we're providing module
URIs in some contexts (such as links to new pages) should
Outline this and all the other examples of cajole
the things that need to be TODOne.
|CAVEAT)\b/g);
' || part === 'CAVEAT') {
We should probably
Given that we maintain the invariant that
Given that we maintain the invariant that
untameRecord isn't defined yet
Can all this JSON stuff be moved out of the TCB?
We may deprecate this in favor of <pre>
Not clear that this is the performant
Should nonextensible objects be stampable?
The only remaining use as of this writing is
Find out if this is still the plan.
The latest draft errata fixes this. We'll be ratifying
Where this is used, do we really want frozenness
Deprecate this API, since it requires that we leave
Clarify the meaning of (a||b)* and modify this function
Is there any value in enumerating elements?
divert IllegalArgumentExceptions to MessageQueue
divert IllegalArgumentExceptions to MessageQueue
divert IllegalArgumentExceptions to MessageQueue
CSS2.1 has changed lexical conventions to effectively
all clients should pass in a proper queue
rewrite with a transition table or just use ANTLR
can this first clause go away?
Once IE allows escapes inside URLs, then
stop using java.net.URI and use a decent URL
maybe reliably throw an exception type, that includes
this should never be rendered since rendering a group
is this true?
maybe enforce the convention that there are matched
check against XML&HTML definitions of whitespace.
do away with the below once Shindig has done away with Neko.
do away with these once shindig has gotten rid of Neko
This class relies on
Investigate whether use of continue to a non loop
Do we want to remove labelling of non-loop statements
Erase the distinction between LabeledStmtWrapper and
audit all places where (instanceof Expression) is used since we have
breaksReaching should probably take a parameter of type
rename breaks and continues to collectBreakTargets and
move unescaping to Escaping.java -- nobody will look there
instead of using a block, introduce a statement collection as a super
Remove the concept of "synthetic" entirely.
skip if the self name is never used.
move this into scope if we use a single CajitaRewriter to rewrite
How does this work? Why not return node?
originalSource
sourceLocationMap
imports
manifest
once decls consolidated, no need to add to
limit
limit further
Limit further
Need a general way to expand lValues
Need a general way to expand lValues
Need a general way to expand readModifyWrite lValues.
better lower limit
better lower bound
Figure out when post increments are being
If we ever turn on taint checking for
check after Kona meeting
Need a general way to expand readModifyWrite lValues.
better lower limit
better lower bound
Figure out when post increments are being
move this into scope if we use a single ES53Rewriter to rewrite
How does this work? Why not return node?
originalSource
sourceLocationMap
imports
manifest
(metaweta) Implement block scoping.
once decls consolidated, no need to add to
(metaweta) Implement immutability auditor and then only allow
limit
Use fastpath
limit further
It's not masking, it's replacing; invent a different error.
Use fastpath.
Use fastpath.
Need a general way to expand lValues
Need a general way to expand lValues
Need a general way to expand readModifyWrite lValues.
better lower limit
better lower bound
Figure out when post increments are being
Do a lighter-weight wrapping when the function
Only use closure if it's really recursive.
use a whitelist, deactivate the rest by replacing with
Add case for dotted path expression.
encode (or at least sanitize) key
to be read in from a JSON config
support getters and setters in object quasis
As a special case, an Identifier with a null value is
consider returning a defensive copy rather than
If we ever have a cheap way to test whether result is used freely
These rules should be independent of whether we're writing
Figure out a way to leave key alone and
Add metadata for a rule indicating that it always fails,
importedVariables is only used by the root-most scope; it
Uses private access to 's' which is of same class but distinct
Uses private access to 's' which is of same class but distinct
Refactor to use standard Caja Visitor. Currently not
Change the ParseTreeNode type for the right hand sides
Permit should generate a JSON AST directly,
can this move from the superclass into a sibling class?
Need to pass in the URI rewriter from the build
limit to JSON
make this subclassable so opensocial specific flags can be
remove once JS uri policy is okay
check argument if child now a FunctionCall
for content: and other URI types, use
this is currently 6 instead of 4 because it also limits
propagate error message
should we whitelist nodes here, to e.g. prevent
Safari 2's defaultView wasn't a window object :(
can bubbling and cancelable on events be simulated
when we change event dispatching to happen
How do we emulate 'useCapture' here?
How do we emulate 'useCapture' here?
Put a class on this so if the host page cares about 'all iframes'
justify this set of refs as being sufficient to do
Our implementation of getAttribute (and friends)
Come up with a uniform convention (and helper functions,
Improve implementation (interleaving, memory leaks)
Do we need more attributes of the event than 'target'?
send()-ing an empty string because send() with no
Expect tamed XML document; unwrap and send
Implement a taming layer for XML. Requires generalizing
Does this work on IE, where console output
integrate cajita compiler to allow arbitrary
determine mime type properly.
make sure it really is a DOM node
upward navigation breaks capability discipline.
Come up with some notion of a keeper chain so we can
Add code to list all the other handled stuff we know
If they are not under the same virtual doc root, return
Add code to list all the other handled stuff we know
make nodeClasses work.
Due to the following bug:
make sure event doesn't propagate to dispatched
make sure event doesn't propagate to dispatched
Add code to list all the other handled stuff we know
create a proper class for BODY, HEAD, and HTML along
If they are not under the same virtual doc root, return
maybe whitelist the color names defined for CSS if the arg is a
remove these, and only expose them via window once
figure out a mechanism by which the container can
Implement in terms of
expose a read-only version of the document
need a testcase for this
Implement
define on prototype.
Build a more sophisticated virtual class hierarchy by
the above probably does the wrong thing in the case where
allow subbing module id resolver
Using XHR loader didn't work; why?
This needs to be redefined in terms of effect/loader;
review error handling -- are errors propagated and descriptive?
does not support dependencies. Needs to tie in to
do we want to reject multiple attach attempts?
We could append the cajoled HTML to existing contents of the
π is different from Π
validate sanitizer regexs against the HTML5 grammar at
relying on sanitizeAttributes not to
Nothing is gained by having this be a partial override of
We do *not* want to support multiple HTML files
http://code.google.com/p/google-caja/issues/detail?id=994
http://code.google.com/p/google-caja/issues/detail?id=994
Once I have network access, figure out if this is here because of a mis-merge?
this could probably be more simply done as a
output to a message queue
this is problematic for DOM nodes parsed without
build up a list of classes and ids for use in
Will add UncajoledModule wrapper when we no longer
do we need to emit a static attribute when the
can we get rid of this noop by getting rid of the
Investigate why this is the right criterion to use.
rename this to CommonMessageType and rename
Change CajaArguments to handle >1 occurrence of arg
The service like the gwt version should accumulate
this is not a valid assumption.
Use a standard class or move to appropriate package
Fix the "unrecoverable" error responses throughout
Would a {@link com.google.caja.reporting.MessageType} be
replace "may be null" and "not null" shorthands with
replace with com.google.common.Predicate and the
Once EcmaScript and Valija allow patterns in parameter
If the resolution of bug #814 is for
figure out why things break when the
Investigate why things break if the following line
Why are we testing for the empty string here?
fix this once DONTENUM properties are better
the positions below seem to be wrong
should work if there is a break here.
make an ASCII-art Rhino.
Could optimize out foo.
better file positions for error messages.
Find a nice collection of "typical" html files!
cajole the scripts so that for loops work. -->
cajole the scripts so that for loops work. -->
cajole the scripts so that for loops work. -->
- do we need a special test for this? is x2v working"
- wait on MarkM to figure out what\'s up with this test."
not sure about this test
- not sure about this one -->",
- I don\'t think attendee stuff is implemented in X2V"
- should \'REQ-PARTICIPANT be lc\'ed? -->",
- should mailto: be here? -->",
- ask MarkM about this. Did he mean to use <del>? -->",
- review the x2v formatting stuff for DESCRIPTION"
-->",
- eh? -->",
cajole the scripts so that for loops work. -->
implement hourly iteration
implement minutely iteration
implement minutely iteration
is this right?
check advancement of more examples
cajole the scripts so that for loops work. -->
cajole the scripts so that for loops work. -->
check that these answers are right
test getSymbol
Should whitespace after the units in quantities be escaped
test rendering of @imports, @page, @font-face,
test rendering of unicode range literals.
*/
*/
*/
should be weakly keyed, but only used for short lived tests.
this error message seems to be a bug.
Why is this here?
These tests were generated by the fuzzer - distill them to
better comment each of the test input files.
Enhance test framework to allow "before" and "after"
This test isn't green because we haven't yet fixed the bug.
Need more tests.
The below should throw MessageType.MASKING_SYMBOL at
Once permit templates can be loaded dynamically, create
The below test is not as complete as it should be
SECURITY: Re-enable by reading (say) x.foo, and
Fix when bug 956 is fixed.
Refactor the test cases so that we can use CajitaModuleRewriter
failure should no longer be an option on (S)ES5/3.
failure should no longer be an option on (S)ES5/3.
failure should no longer be an option on (S)ES5/3.
failure should no longer be an option on (S)ES5/3.
SECURITY: Re-enable by reading (say) x.foo, and
Fix when bug 953 is fixed.
Fix when bug 953 is fixed.
Fix these tests to test for the output we now expect.
Fix when bug 953 is fixed.
Need more tests.
Refactor tests to use checkAddsMessage(...) instead
Change dependents to use checkAddsMessage and
Change checkAddsMessage and similar functions to check
deploy the already-configured war instead of manually
print a warning here?
fix message. "fixed" is well-formed but disallowed.
We could break the position rule into multiple
Figure out why RhinoTestBed.runJsUnittestFromHtml isn't
Move as many of these as possible to
Once they decide on scoping & initialization rules, test
async requirements are not counted in the test status.
rewrite XHR and setTimeout tests to use this scheme.
write this properly.
enable for all node-types once EQ fixed on IE
enable for all node-types once EQ fixed on IE
Add negative tests when virtual hierarchy is improved:
webdriver doesn't have a good way of testing relatedTarget.
This is wrong because we don't allow the insertion point
This test should be replaced by ones for the proper
this test often causes ie6/ie7 to crash. figure out why
test that getElementsByTagName and getElementsByClassName
also need to test when onload isn't set
can't verify focus changes in webdriver, because webdriver runs
arrange to reliably detect loading errors in loadCaja and report
do this with nice output without using jsUnitCore internals?
When the whole URI-policy-implementation thing is cleaned
create jsunitRegisterAsync(), then make passes implicit.
move parameters into a separate namespace so they
Change tests to use structural equality (via quasi
maybe actually diff using
maybe actually diff using
maybe replace this with the JSR 223 stuff.