|
History
| Log In
|
|
|
|
[
Permlink
| « Hide
]
Henry Minsky - 03/Jun/06 08:17 PM
This is a test case from Jake Yang. Note, I had to recompile the java servlet classes to make it load into Tomcat
The LzLoader returnData method got rewritten when we switched to client-parsed data, and was stripping the header data from the server response, for no good reason.
I have a patch pending to fix this. The patch was in returnData: LzLoader.prototype.returnData = function ( loadobj , data ){ // Check if returnData has already been called on this // object. This can happen if a serverless data load timed out in // the LFC, but eventually returned something via the // LoadVars.sendAndLoad() callback. if (loadobj.loaded) { if ($debug) { Debug.warn("%w.returnData: %w already loaded", this, loadobj); } return; } if (loadobj instanceof MovieClip) { // Nothing to do } else if (loadobj instanceof XMLNode) { // Do a fixup to add one extra level of data node to the data, // to be compatible with what the DataCompiler produces. var ndata = new LzDataElement('resultset'); // The raw XML response string was stashed here by LzLoadQueue.serverlessOnDataHandler // TODO [2005-08-11 ptw] Is there not an issue that there // could be multiple requests? if (this['rawtext']) { ndata.rawtext = this.rawtext; } if (typeof(data) != 'undefined') { if (loadobj.proxied) { // proxied req gets back <resultset><body>XML</body><headers/></resultset> ndata.setChildNodes(data.childNodes[0].childNodes); ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ } else { // SOLO, we get back raw XML only ndata.setChildNodes([data]); } } data = ndata; } else { if ($debug) { Debug.error("%w.returnData: unhandled loadobj %w", this, loadobj); } } // Ok, we have loaded this. loadFinished changes state of loadobj // from loading to loaded. // TODO [2005-08-11 ptw] If this were a multi-threaded system, we // would have to worry about getLoadMovie trying to reuse the // loadobj before we have finished with it. // [2005-08-25 ptw] So, we don't want to mark the clip as loaded // if this is a persistent connection? Not sure what that means. if ( ! this.doPersist ) _root.LzLoadQueue.loadFinished( loadobj ); // [2005-08-25 ptw] Is this correct? Do we really want an aborted // request to signal that it timed out? if ( ! loadobj.valid ) { this.doTimeOut(loadobj); return; } this.lastloadtime = getTimer() - loadobj.loadtime; var err; if ( data.childNodes[0].nodeName == "error") { err = data.childNodes[0].attributes.msg; } else if (data.nodeName == "error") { err = data.attributes.msg; } else if (data == "medialoaderror") { err="Media load error"; } if ( err != null ){ Debug.error( err ); this.onerror.sendEvent( data ); return; } if ( this.queuing ){ loadobj.data = data; while ( this.expectedList.length && this.expectedList[ 0 ].loaded ){ var tlmc = this.expectedList.shift(); if ( tlmc.timedout ){ this.doTimeOut(tlmc); } else if( tlmc.data ){ this.ondata.sendEvent( tlmc.data ); tlmc.data = null; } } } else { if ( loadobj.timedout ){ this.doTimeOut(loadobj); } else if( data ){ this.ondata.sendEvent( data ); } } } Note, we also need to take changeset listed below:
I made a fix, changeset 40979 on 2006/03/29 which fix the set-header behavior. I am wondering now if that made it into the 3.2 (or 3.3) releases. Jake, can you tell me the value that prints out in the debugger if you type LzLoadQueue.qargs ? It should be this list. If "headers" is missing ,that is the bug, you can add this code as a patch. LzLoadQueue.qargs = [ 'nsprefix', 'trimwhitespace', 'sendheaders', 'headers', 'reqtype', 'timeout', 'fpv', 'ccache']; Now that this is a 3.4 bug, I'll let Henry integrate his patch into svn openlaszlo/trunk. Assigning to Henry.
I integrated this change and added a test case to test/lfc/data/alldata.lzx
|
|||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
 |
|
Bug
Resolved
P2
response headers dont' show up anymore in datasets
