[OpenLayers-Users] Re: SOLVED - Firefox nsIXMLHttpRequest.send throwing 0xc1f30001 (NS_ERROR_NOT_INITIALIZED)

[Andrey]

On 05/11/2011 12:30 PM, Andrey wrote:
> On 05/11/2011 08:07 AM, Eric Lemoine wrote:
>> On Tuesday, May 10, 2011, flexer<Andrey.Aniskovets at gmail.com>  wrote:
>>> I have got the same error (Exception... "Component returned failure 
>>> code:
>>> 0xc1f30001 .... )
>>> And I have found another solution:
>>>
>>> Do NOT call "this._object.send(vData);"
>>> if  "this._object.abort();" has been called (see XMLHttpRequest.js).
>>>
>>> -----------------------
>>> Look at Request.js:
>>>              window.setTimeout(function(){
>>>                  if (request._aborted !== true) {
>>>                      request.send(config.data);
>>>                  }
>>>              }, 0);
>>>
>>> and look at XMLHttpRequest.js :
>>>      cXMLHttpRequest.prototype.send    = function(vData) {
>>>          // Add method sniffer
>>>          if (cXMLHttpRequest.onsend)
>>>      ....
>>>
>>> and look at XMLHttpRequest.js:
>>>     cXMLHttpRequest.prototype.abort    = function() {
>>>          ....
>>>          // BUGFIX: Gecko - unnecessary DONE when aborting
>>>          if (this.readyState>  cXMLHttpRequest.UNSENT)
>>>              this._aborted    = true;
>>>          this._object.abort();
>>>          ....
>>> -----------------------
>>>
>>> I have added to XMLHttpRequest.js these lines and the problem was 
>>> solved:
>>>
>>>      cXMLHttpRequest.prototype.send    = function(vData) {
>>>          if (!!this._isAbortCalled)
>>>              return;
>>>          ....
>>>
>>>      cXMLHttpRequest.prototype.abort    = function() {
>>>          this._isAbortCalled    = true;
>>>          ....
>>>
>>> ---------------------
>>> I.e.
>>>        try { this._object.send(vData); } catch (e) {}
>>> looks like a workaround.
>>> What do you think?
>>> Could you check my changes on your system if it solves your problem?
>> Hi
>>
>> Is the problem reproducible on one of the OpenLayers examples?
>> (http://www.openlayers.org/dev/examples)
>>
>> What version of OpenLayers do you use? Do you have the problem with
>> OpenLayers trunk?
>>
>> XMLHttpRequest.js already includes a state variable to know if abort()
>> has been called, namely _aborted. But abort() sets _aborted to true
>> only if the XHR status is greater than UNSET. Could you try to have
>> abort() unconditionally set _aborted to true and see if that fixes it?
>> Thanks.
>>
> Hi
> -------
> It looks like the problem is reproducible here : 
> http://www.openlayers.org/dev/examples/getfeature-wfs.html  :
> POST 
> http://www.openlayers.org/dev/examples/proxy.cgi...p%3A%2F%2Fv2.suite.opengeo.org%2Fgeoserver%2Fows
> POST 
> http://www.openlayers.org/dev/examples/proxy.cgi?url=http%3A%2F%2Fv2.suite.opengeo.org%2Fgeoserver%2Fows
>
> Aborted
>
> OpenLayers.js (line 969)
>
> - that is from Firebug console.
> - there is ".send..." method;
>  then I have manually wrap "this._object.send(this._data);" with  
> "try{..}catch(e){alert(e)}"  and have got this exception :
> [Exception... "Access to restricted URI denied"  code: "1012" 
> nsresult: "0x805303f4 (NS_ERROR_DOM_BAD_URI)"  location: 
> "file:///media/TMP/tmp_shara/OL-ex/getfeature-wfs_files/OpenLayers.js 
> Line: 11177"]
> So seems I have done something wrong.
> -------
> We use 2.10 (stable) version of OpanLayer.
> -------
> The problem occurred right in the case XHR status is NOT greater then 
> UNSENT (on abort() call).
>
> I have removed my fix and have commented this line:
>    if (this.readyState > cXMLHttpRequest.UNSENT)
> - this also fixes the problem: checked on FF3.6.17+ubuntu; 
> F3.6.17+winXP; IE8+winXP ;
> - BUT that line is marked as "BUGFIX" !
>  - I have also found a strange side effect at IE (! but NOT at FF) 
> (after commenting that line) : ... but seems that is completely 
> another problem and most probably it does NOT relate to current 
> ticket: "this.blocks.length" is null or not an object Framed.js, 
> line:140 character 24
>
> Thanks.
Hi,
please ignore my comment about "strange side effect" - that definitely 
does NOT relates to current ticket - I have reproduced it without 
commenting bugfix for "Gecko".
Thanks.