Bugzilla – Full Text Bug Listing
| Summary: | [web connector] "RSS Error" trying to connect to in house ticket system using Web Templates | ||||||||
|---|---|---|---|---|---|---|---|---|---|
| Product: | z_Archived | Reporter: | Jesse Rosalia <theJenix> | ||||||
| Component: | Mylyn | Assignee: | Project Inbox <mylyn-triaged> | ||||||
| Status: | CLOSED MOVED | QA Contact: | |||||||
| Severity: | normal | ||||||||
| Priority: | P3 | CC: | steffen.pingel | ||||||
| Version: | 3.0 | Keywords: | helpwanted | ||||||
| Target Milestone: | --- | ||||||||
| Hardware: | PC | ||||||||
| OS: | Windows XP | ||||||||
| Whiteboard: | |||||||||
| Attachments: |
|
||||||||
Created attachment 116995 [details]
A screen cap of the error message
Incidentally, I would love to help take a look at this further but I couldn't figure out where to get the source from. The CVS link I did find (:pserver:anonymous@dev.eclipse.org:/cvsroot/tools/org.eclipse.mylyn) kept asking for a password and the web view didn't seem to have any relevant files in the org.eclipse.mylyn.web.* packages. The web connector is in the org.eclipse.mylyn.web.tasks plug-in. Make sure you use :pserver:anonymous@dev.eclipse.org:/cvsroot/tools as your cvs root. Check out instructions are here: http://wiki.eclipse.org/Mylyn_Contributor_Reference#Checkout (In reply to comment #3) > The web connector is in the org.eclipse.mylyn.web.tasks plug-in. Make sure you > use :pserver:anonymous@dev.eclipse.org:/cvsroot/tools as your cvs root. Check > out instructions are here: > > http://wiki.eclipse.org/Mylyn_Contributor_Reference#Checkout > Steffen, Thanks for the reply...I thought I tried CVSROOT that but apparently I didnt as its now working. I did see that page earlier, and I couldn't get the .psf file import to work (same problem I was having earlier with the command line...kept asking me for a password). Anyway, I have the source now...Ill refer to that link for how to test, etc...if anyone has any idea based on the description what the problem could be or where I should start looking, please let me know. -Jesse An update...I found the cause of the problem I described and a workaround but I think its worth some time to examine the code that was tripping me up. In a nutshell, the problem was in WebRepositoryConnector#performQuery, where the following code exists:
String taskPrefixAttribute = query.getAttribute(KEY_TASK_PREFIX);
if (!Util.isPresent(taskPrefixAttribute)) {
return performRssQuery(content, monitor, resultCollector, repository);
} else {
String taskPrefix = evaluateParams(taskPrefixAttribute, queryParameters, repository);
String queryPattern = evaluateParams(query.getAttribute(KEY_QUERY_PATTERN), queryParameters, repository);
return performQuery(content, queryPattern, taskPrefix, monitor, resultCollector, repository);
}
I had not set a task url for my task repository, so this code was treating it like an RSS feed. While I was able to overcome this problem by providing a task url, it seems arbitrary to switch between a RSS and patterned query based on this attribute. I noticed the following code in WebQueryWizardPage UpdatePreviewJob#run:
if (queryPattern != null && queryPattern.trim().length() > 0) {
status = WebRepositoryConnector.performQuery(webPage, queryPattern, taskPrefix, monitor, collector, getTaskRepository());
} else {
status = WebRepositoryConnector.performRssQuery(webPage, monitor, collector,getTaskRepository());
}
This makes a bit more sense to me, checking to see if a query pattern has been provided as a means of determining how to query the data source.
I'd like a bit of advice on how to proceed here...is it appropriate to turn this bug into an enhancement request for the purposes of evaluating this logic? Should I make (and test) the chance and submit it as a patch for review?
Eugene, you wrote the original code. Can you recommend how to proceed here? (In reply to comment #6) Looking at CVS history it seem like regression introduced by your changes - http://dev.eclipse.org/viewcvs/index.cgi/org.eclipse.mylyn/org.eclipse.mylyn.web.tasks/src/org/eclipse/mylyn/internal/web/tasks/WebRepositoryConnector.java?root=Tools_Project&view=diff&r1=1.51&r2=1.52 An empty "pattern" value been used as an indicator to use rss to parse results and I believe this is explained on the wiki. Also, when rss mode is enabled, web tasks need to be created differently (id, url and descriptions). An outsiders opinion here...I think having the query aware of what type of query it is makes sense (the isRss flag). Or perhaps taken a step further, maybe the query object should have specific performQuery method, which would be implemented one way for a RssQuery type and another way for a PatternableQuery type (move the logic from the 2 static methods in WebRepositoryConnector). That way the check to determine what type of query it is only needs to be performed in one spot...at the time of query definition/creation...the code that operates on that query later then doesnt need to worry about it. For curiosity's sake, I tried using if (queryPattern != null ...) conditional in WebRepositoryConnector#performQuery, and then blanked out my task url. It appears to work except for some reason the task item strings no longer contain the task id. Any ideas where that name/string gets built? Mylyn has been restructured, and our issue tracking has moved to GitHub [1]. We are closing ~14K Bugzilla issues to give the new team a fresh start. If you feel that this issue is still relevant, please create a new one on GitHub. [1] https://github.com/orgs/eclipse-mylyn |
Created attachment 116994 [details] A sanitized capture of the query results (originally pulled using the "Open" from inside the advanced config in the new query dialog. Build ID: M20080911-1700 Steps To Reproduce: 1. Create a task repository to system. Use the following query pattern: <a class="ticketLinks" href=".*">({Id}.+?)</a></td> <td class="searchResults"> ({Owner}.+?)</td> <td class="searchResults"> ({Status}.+?)</td> <td class="searchResults"> .*</td> <td class="searchResults"> ({Type}.+?)</td> <td class="searchResults"> ({Description}.+?)</td> 2. Create a query that ensures the results back in that format. 2a. Preview (inside the new/modify query screen) returns correct results. 3. Save the query. Eclipse will attempt to query the task repository and will respond with the following error: "Failed to parse RSS feed: "Invalid XML: Error on line 30: The entity name must immediately follow the ", in the entity reference." More information: Ive attached a sanitized (removed company proprietary info) capture of the output, pulled by using the "Open" button next to the Query URL box on the new query dialog. I think the problem may have to do with javascript in the web query results...line 30 of that file: if(window.opener != null && strSavedQuery == "null"){ Ive also seen a similar error on the same line complaining about the &.