Persona is no longer an option for authentication on BMO. For more details see Persona Deprecated.
Last Comment Bug 606818 - RSS Subscription stuck at "verifying the feed"
: RSS Subscription stuck at "verifying the feed"
Product: MailNews Core
Classification: Components
Component: Feed Reader (show other bugs)
: unspecified
: x86 Windows 7
: -- major with 1 vote (vote)
: Thunderbird 11.0
Assigned To: alta88
: 555789 (view as bug list)
Depends on:
  Show dependency treegraph
Reported: 2010-10-24 09:12 PDT by marvinhk
Modified: 2011-12-13 05:58 PST (History)
5 users (show)
See Also:
Crash Signature:
QA Whiteboard:
Iteration: ---
Points: ---

Illustation of a workaround (2.04 KB, patch)
2011-02-12 01:24 PST, andershol
no flags Details | Diff | Splinter Review
Updated workaround (1.28 KB, text/plain)
2011-09-28 15:09 PDT, andershol
no flags Details
handle throw. (1.07 KB, patch)
2011-12-02 14:16 PST, alta88
no flags Details | Diff | Splinter Review
update. (752 bytes, patch)
2011-12-05 07:38 PST, alta88
neil: review+
Details | Diff | Splinter Review
new patch. (1.09 KB, patch)
2011-12-06 09:58 PST, alta88
no flags Details | Diff | Splinter Review
new patch. (1.04 KB, patch)
2011-12-07 09:14 PST, alta88
neil: review+
Details | Diff | Splinter Review
patch. (1.01 KB, patch)
2011-12-09 08:29 PST, alta88
alta88: review+
standard8: approval‑comm‑aurora+
Details | Diff | Splinter Review

Description marvinhk 2010-10-24 09:12:52 PDT
User-Agent:       Mozilla/5.0 (Windows; U; Windows NT 6.1; en-GB; rv: Gecko/20101012 Firefox/3.6.11 (.NET CLR 3.5.30729)
Build Identifier: Mozilla/5.0 (Windows; U; Windows NT 6.1; en-GB; rv: Gecko/20101013 Thunderbird/3.1.5

Subscription to RSS feeds from the Yahoo Blog always stop at "verifying the feed"

Reproducible: Always

Steps to Reproduce:
1. Right click an account and choose Subscribe
2. Add feed URLs from Yahoo Blog (e.g.
3. TB keeps verifying the feed without completion
Actual Results:  
TB keeps verifying the feed without completion

Expected Results:  
RSS subscribed to the corresponding folders of the account.
Comment 1 andershol 2011-02-12 01:24:07 PST
Created attachment 511932 [details] [diff] [review]
Illustation of a workaround

I had this problem with the Planet Postgresql feed ( ). Looking in Thunderbirds error console, I found this:

Error: [Exception... "Component returned failure code: 0x804b000a (NS_ERROR_MALFORMED_URI) [nsIStandardURL.init]"  nsresult: "0x804b000a (NS_ERROR_MALFORMED_URI)"  location: "JS frame :: chrome://messenger-newsblog/content/FeedItem.js :: anonymous :: line 58"  data: no]
Source File: chrome://messenger-newsblog/content/FeedItem.js
Line: 58

As a simple fix, I tried wrapping the setter found there in a try/catch. This enabled Thunerbird to get past the Verifying-state and into "Importing feed" (I think it was called), but then it stopped half way through with this in the error console:

Error: s is null
Source File: chrome://messenger-newsblog/content/utils.js
Line: 395

Treating null as the empty string allowed to feed to be imported.

The attached patch illustrates the changes I made to get the feed imported, but I don't known if this is the correct way to solve the problem.
Comment 2 Ludovic Hirlimann [:Usul] 2011-02-18 06:35:24 PST
Hum asking on irc in #maildev is your best option (try to find myk and ask him).
Comment 3 Fernando Hartmann 2011-03-02 10:23:43 PST
The happens to me in 

Mozilla/5.0 (Windows; U; Windows NT 6.1; en-US; rv: Gecko/20110221 Thunderbird/3.1.8
Comment 4 andershol 2011-09-28 15:09:25 PDT
Created attachment 563192 [details]
Updated workaround

The workaround I now use is illustrated in this patch (unzip omni.jar, make the change and re-zip omni.jar). Instead of clearing the url, it just appends "#badurl" to the problematic urls to ease debugging.
Comment 5 alta88 2011-12-02 14:16:18 PST
Created attachment 578721 [details] [diff] [review]
handle throw.
Comment 6 alta88 2011-12-02 14:33:55 PST
*** Bug 555789 has been marked as a duplicate of this bug. ***
Comment 7 2011-12-02 16:31:53 PST
Comment on attachment 578721 [details] [diff] [review]
handle throw.

I had a look to see what these invalid URIs were, and they were tag: URIs. Perhaps we should be using the IO service to parse these URIs instead of hoping that they are standard URLs?
Comment 8 alta88 2011-12-02 17:16:06 PST
it seems many uris are standard, but certainly not all.  my assumption here is, if they are standard then process that way, if not then just use what the publisher gave (or how current feed processing assigns an item url based on what the publisher gave).

i don't think it matters much for the feed system - they are used as messageIds and keys into feeditems.rdf.  the critical thing is to handle this (sort of bogus) error and mush on cleanly.
Comment 9 2011-12-04 05:04:56 PST
(In reply to alta88 from comment #8)
> i don't think it matters much for the feed system - they are used as
> messageIds and keys into feeditems.rdf.
Then I don't see the point of trying to mash them into a standard URL...
Comment 10 andershol 2011-12-04 13:40:13 PST
If you are going to fall back on just using the value as is, if it can't be parsed as an url, why attempt to parse it as an url at all? why not just use the value?

At least for the feed, that I have problems with, firefox (specifically the code that shows the code when you browse to the url of the feed) seems to parse it correctly, so it might be helpful to compare the feed handling from the two applications. It turns out that the feed have an error in that the isPermaLink-attribute on the guid-element is set to "true" even though the guid-element does not contain a valid url.

If I interpret the code correctly and am looking in the right place, it seems firefox first look at the link-element of the feed entry, and then only if it is emty, look that the guid-element. Also it compares isPermaLink-attribute case-insensitively with "false". Therefor it is not tripped by the error as the link-element in my case contains a url and so it do not use the guid for the url.

The current thunderbird-code on the other hand, seems to use the value of the guid-element if the isPermaLink-attribute is missing or (case-sensitively) "true" and only check the format if attrbiute is missing, and else use the link-element.

So if the value of the guid-element, with the isPermaLink-attribute set to true, should continue to be preferred over the link-element, perhaps line 145 of feed-parser.js could just be changed from "if (!guidNode.hasAttribute("isPermaLink"))" to "if (isPermaLink)" to have the check be preformed in all cases where the guid-element is used. That just leaves the value unchecked if the value of the link-element is used.
Comment 11 alta88 2011-12-05 07:38:32 PST
Created attachment 579064 [details] [diff] [review]
Comment 12 alta88 2011-12-05 08:20:39 PST
(In reply to andershol from comment #10)

a truly noble endeavor would be bug 450543.  but it's not clear the toolkit code is issue free, even per its author:

to your specific guid/permalink question - i believe the way Tb handles isPermaLink is correct according to the spec:

the solution of what to do with a guid that must be a valid url but isn't, will be the same - ensure no halt in processing, do a best effort to verify the url, but in the end if the publisher gave you non spec junk what can you do but store it and not assume anything.

so the new patch stores as is, though i lean to the fact that there is a non zero positive in making a standard url if it was meant to be a url.
Comment 13 2011-12-06 02:01:21 PST
(In reply to alta88 from comment #12)
> so the new patch stores as is, though i lean to the fact that there is a non
> zero positive in making a standard url if it was meant to be a url.
But I still think running URLs through the nsIOService is better than directly creating a nsStandardURL object.
Comment 14 alta88 2011-12-06 09:58:02 PST
Created attachment 579349 [details] [diff] [review]
new patch.

i completely agree, and mistook your prior comment to be 'save as is'.
Comment 15 2011-12-07 07:26:15 PST
Comment on attachment 579349 [details] [diff] [review]
new patch.

>-    var uri = uri.QueryInterface(Components.interfaces.nsIURI);
>+      url = ioService.newURI(aVal, null, null)
>+                     .QueryInterface(Components.interfaces.nsIURL);
I don't know why you thought you had to query to nsIURL; nsIURI was used above to get access to .spec but newURI already returns an nsIURI so you can access the spec straight away.

>+    this.mURL = url ? url.spec : aVal;
This could probably be simplified by using separate assignment statements inside the try and catch blocks respectively.
Comment 16 alta88 2011-12-07 09:14:50 PST
Created attachment 579718 [details] [diff] [review]
new patch.
Comment 17 2011-12-07 12:17:02 PST
Comment on attachment 579718 [details] [diff] [review]
new patch.

>+    var url;
>+      url = ioService.newURI(aVal, null, null);
>+      this.mURL = url.spec;
(It's probably not worth having a separate variable for this.)
Comment 18 alta88 2011-12-09 08:29:29 PST
Created attachment 580424 [details] [diff] [review]
Comment 19 Mark Banner (:standard8) 2011-12-13 05:34:18 PST
Comment on attachment 580424 [details] [diff] [review]

a=Standard8 for aurora. Given where we are for beta, I don't really want to take this this late - we'll pick it up in 10.
Comment 20 Mark Banner (:standard8) 2011-12-13 05:58:55 PST
Checked into trunk and aurora:

Note You need to log in before you can comment on or make changes to this bug.