Apache OpenOffice (AOO) Bugzilla – Issue 23786
Unable to open documents hosted by a certain web application
Last modified: 2013-02-07 22:12:35 UTC
Hello, Sorry if this is the wrong place to post this - I spent ages trying to figure out where to put it! (if I could figure out where to post that comment I would do it separateley - finding this very difficult to use!). It appears that openOffice cannot view documents which are contained in a secure server. I have attempted to access documents stored in WebCT (a web- based e-learning application; uses ticket-based authentication). When I attempt to open such a document within the browser (tested with MS PowerPoint document) I receive a Forbidden Error from OpenOffice. Early versions MS Office XP apparently had a similar problem (now fixed) which is documented here: http://support.microsoft.com/default.aspx?scid=kb;en- us;Q314535 - this url may help explain the problem better than me! Here is the error message displayed by OpenOffice: Error Loading Document http://webct1.brunel.ac.uk/MG2113/intro.ppt Could not establish Internet Connection to http://webct1.brunel.ac.uk Please feel free to contact me if you have any queries. I should be able to organise you access to the abovementioned WebCT server for testing purposes if necessary. Regards, James Fraser
Kai, I think this is something for you.
Hmm, according to the error message OO is not able to connect to the server named webct1.brunel.ac.uk . It even does not try to load the document, because this requires a server connection. Have you checked that your HTTP-Proxy settings are correct (OO: Tools->Options->Internet->Proxy)? Do you need to use a proxy to communicate with that server or, the other way around, does using a proxy block the communication? Please check your proxy settings.
Just another thing: Is the content of the error message really 'Could not establish Internet Connection to http://webct1.brunel.ac.uk' or is it actually 'Could not establish Internet Connection to webct1.brunel.ac.uk'
indigofour, cannot reproduce the problem. Seems to be a configuration error (HTTP proxy?). Host cannot be reached. In case the proxy configuration is not the problem, feel free to hand issue back to me and provide me access to the server in question, please.
Responding to your questions: Mon Jan 5 01:06:13 -0800 2004: I do not use a proxy to connect to the internet. However, my ISP may do (would that affect it?). Mon Jan 5 01:09:46 -0800 2004: I just checked again and I am getting a different message! It is as below (for a different document as it happens - but I checked several - couldn't remember where I got the original one from!) "Error loading document http://webct1.brunel.ac.uk/web- ct/courses/EE2101A/McLibel_Trial.ppt: Error reading data from the Internet Server error message: 403 Forbidden." As you can see however, the 'http' is there. FYI: I use ZoneAlarm firewall, but disabled this fully to ensure that wasn't the problem. As I am involved with the adminstration of this particular server (as well as being a user), I can give you a user account. Please contact me directly by email on james.fraser@brunel.ac.uk so I can arrange this (wasn't sure if kso@openoffice.org is a real email address or a group list?). I can also pass any questions you may have for the people who do all the complicated stuff with the server for us. Thanks for your help, James.
Re-assigning to kso - I think that's what I was supposed to do...
James, I login to WebCT and go to the page containing the ppt and doc. After clicking on any of the two links, the browser downloads the document and starts OOo passing the URL to the downloaded document as an argument. OOo loads the document just fine. No problems. If I just start OOo and enter one of the problematic URLs you mentioned above, for example http://webct1.brunel.ac.uk/web-ct/courses/EE2101A/McLibel_Trial.ppt, OOo displays an error message, exactly like you reported. To go ahead, I need to know where the problematic URLs you mention above come from. Do you copy them somewhere out of the WebCT application? I noticed that the working URLs behind the document links in the web application look very differrent from what you're using Example: javascript:top.update_loc_to_new_tool('PowerPoint (pages/powerpoint.ppt)', 'PowerPoint (pages/powerpoint.ppt)', '/SCRIPT/linkalarm/scripts/student/serve_page.pl?1023740511+pages/powerpoint.ppt', '', 3, 0).
James, no response for about a week... I need the requested information to be able to confirm the issue. Please provide the missing pieces of data. In case this is no longer an issue for you, please close this task.
Apologies for the delay in replying - I was expecting to receive an email when new notes were added - I only got the one on the 20th (today). I am following the links in the web-base as usual - not copying or pasting anything. When I click either of the documents from the "Lecture Notes" page, it loads the action menu at the top (the bar of options - part of WebCT), and the main screen goes grey and sits there for 30-40 seconds, before bringing up the error message as before. When I click OK on that, I get a message saying "An API Call exited abnormally" - not sure whether that is generated by OOo or not. I have just tested it using your logon (student account). I am somewhat puzzled that it doesn't affect you too! Perhaps it is, as you suggested earlier, to do with my configuration. I'm not sure where to go from here - please feel free to ask me any relevant questions about my system setup and applications and I will answer as best I can. I will (when I get a chance - mid-exams at the moment!) try installing OOo on my other machine here and see if that has the same problem. James PS: The javascript link that you quoted is obviously something specific to WebCT in the way it serves the files. The difference in the full URLs given in the error message is a result of different logon accounts. Previosuly, when I tested it, I was logged in as a designer - which gives the urls containing just "/<web-base name>/<file path and name>". As a student (like your account), the URL is like: "/web-ct/courses/<web-base name>/<file path and name>", even though these relate to the same file.
James, just checked again. No problems. Your test documents get loaded and displayed okay by my OOo. I still think that you either have a configuration or installation problem. Reinstalling OOo might be worth a try, too. Does this fix your problems? It would be great if you could provide screen shots of the error message box and the crash message box and of your "Tools->Options->Internet->Proxy" configuration page. Following things seem to be sure at the moment: There is no authentication problem. The problem is not related to the Office XP problem you mentioned. I changed the subject of the issue to reflect this.
Changed component to 'ucb', subcomponent to 'code'. Removed issues@framework from CC
No response for more than two weeks now. Cannot confirm task without more information, sorry. Closing task according to our task handling policy.
Hello again, This same issue has now been raised by a student at Brunel, and I'm trying to assist him in my capacity as technical support for the system. Dave (the student) has updated OO to the latest version, checked his browser version is up-to-date etc, but the problem still occurs. Here's some of the information about his PC: Evesham Axis with an Athlon XP 2400 processor - purchased April 2003. Operating system - Windows XP (Sevice Pack 2 update recently installed) Internet explorer 6 Open Office 1.1.0 ISP - Onetel 56k modem AVG antivirus I've reactivated the account I created on WebCT for you so that you can check the specific document that's causing a problem. Here's instructions for what to do: go to http://www.brunel.ac.uk/webbased-learning click Login to WebCT and type your login details (see below) click Research Methods (HH5004) click Lecture Notes click Wk3 readings/resources This loads an MS Word document and causes an error in OO (screenshot will be attached). The username I setup for you is "openoffice" - if you're unsure of the password please email me on james.fraser@brunel.ac.uk There's clearly some problem here, but if it is down to software configuration then it sems a common mistake - as two people have now had the same problem. I look forward to hearing from you.
Created attachment 18612 [details] Screenshot of problem occurring as described on 22nd October 2004, 15:00
This time I'm able to reproduce the problem. Issue confirmed.
Great (sort of). Please let me know if there's any other assistance I (or the student who is experiencing the problem) can provide. When can I expect to hear from you again with further news on this issue? If you need to know how WebCT serves files, I can try and contact the company for you - if so could you give me an appropriate question to ask as I don't know what you'd need to know. James.
Following URL is passed as an argument to the office process that gets started by the browser: http://webct1.brunel.ac.uk/SCRIPT/HH5004/scripts/student/button_bar/HH5004/HH5004wk3furthreading.doc?STATIC+1403765524 One does not need to use the Web application in order to reproduce the problem. Just start soffice and supply the URL above as the first argument.
First, the problem is not related to that special .doc document. It happens with every document stored in WebCT that should be displayed in OOo. It is also not restricted to IE or Windows; it happens on all platforms Second, just discovered: If you directly enter the URL in a freshly started Mozilla/IE, the browser will display the WebCT login page instead of the content of the requested page (the .doc in this case). This is, because you're not logged in to WebCT. The same happens when starting OOo with the URL. The difference between OOo and the browser is that the browser displays the error document the server returns (WebCT login page here) and OOo does not. WebCT uses a cookie to store authentication information. If the cookie is missing (which is always the case in the OOo invocation scenario), the GET request is answered with an error and the error document contains the WebCT login page. OOo does never display those error documents. It always uses dialogs to communicate errors. OOo is no browser! Therefore you do not have the chance to enter your authentication information... and you are lost. Please note, that there are better and standardized ways to implement access to restricted HTTP resources. OOo supports HTTP Basic and Digest authentication mechanisms. If WebCT would use one of these standard ways it would be able to work together with OOo. Sorry, but currently WebCT and OOo are simply not compatible. Implementation of displaying error pages in OOo is real big effort. We will not be able do this in OOo 2.0 time frame. Retargeting issue to OOo later, setting priority to 4.
Adjusted subject.
Andreas, this one is tricky. The problem is that those web apps are implementing their own login user interface. Upon first try to access the protected resource one will be redirected to a login page and the return from the request is that login page along with some error code. OOo never displays error documents. It just interprets the request return code and displays an appropriate error message box. Thus, the user has no chance to enter any login information. The problem described here with the WebCT app goes even one step further. OOo gets started from inside a browser. The browser might even know the login data, because once obtained they are kept in a cookie. Currently there is no way to hand over this cookie to OOo. Even worse, OOo's HTTP client has no cookie support at all.
accepted
ABI->KSO: As discussed ...
.
KSO->TKR: Please take care of this issue.