737966 - Evaluating nsIXMLHttpRequest.responseText throws on certain parsing errors

Status: RESOLVED FIXED

(Core :: DOM: Core & HTML, defect)

Description

[alta88]

Tb feed code does the following:

    this.request = Cc["@mozilla.org/xmlextras/xmlhttprequest;1"]
                   .createInstance(Ci.nsIXMLHttpRequest);
    this.request.onprogress = this.onProgress;
    this.request.open("GET", this.url, true);
    ..
    this.request.overrideMimeType("text/xml");
    this.request.onload = this.onDownloaded;
    this.request.onerror = this.onDownloadError;
    this.request.send(null);

using an invalid feed url, like http://dilbert.com/rss/, returns a parsing error doc in .responseXML, but request.response|responseText throw on evaluation.  this does not happen in Tb11, but does in Tb12 on.  .status is 200.

Error: Component returned failure code: 0x80500001 [nsIXMLHttpRequest.responseText] = <unknown>
Source file: chrome://messenger-newsblog/content/Feed.js
Line: 327

the feed code probably doesn't need to test .responseText and should be asking for responseType=document anyway, but surely the properties should be set to null and not throw.  likely this is a common way to test..

Comment 1

[Boris Zbarsky [:bzbarsky]]

Jonas?

Comment 2

[Masatoshi Kimura [:emk]]

This looks like a regression from bug 590390.
dilbert.com has an invalid charset name "utf-8lias" in response headers.
Before bug 590390, TryChannelCharset updates aCharset only when nsCharsetAlias::GetPreferred has succeeded.
Note that nsCharsetAlias::GetPreferred will truncate the parameter if it failed.
https://hg.mozilla.org/mozilla-central/diff/2893c5e0be5d/content/base/src/nsDocument.cpp
nsXMLHttpRequest::GetResponseText will fail if the document charset is the empty string.

Comment 3

[Boris Zbarsky [:bzbarsky]]

Requesting tracking, since this would be a web-visible regression.

I wonder what the spec says to do here and whether it needs changing....

Comment 4

[Jonas Sicking (:sicking) No longer reading bugmail consistently]

Simply ignoring the channel charset seems like the right thing to do here. That's our previous behavior, right?

Comment 5

[Masatoshi Kimura [:emk]]

We used to return "UTF-8" because document charset was initialized "UTF-8" and no body broke that.
https://mxr.mozilla.org/mozilla-central/source/content/xml/document/src/nsXMLDocument.cpp#177
I think GetPreferred shouldn't break the parameter on failure.

Comment 6

[Makoto Kato [:m_kato]]

Attachment: fix

Comment 7

[Masatoshi Kimura [:emk]]

Could you add a comment explaining why temporary variable (preferred) is required?

Comment 8

[Jonas Sicking (:sicking) No longer reading bugmail consistently]

Comment on attachment 609627 [details] [diff] [review]
fix

Review of attachment 609627 [details] [diff] [review]:
-----------------------------------------------------------------

Though would be great if you could add a mochitest as well to test that things work correctly in a webpage context.

Comment 9

[Makoto Kato [:m_kato]]

https://hg.mozilla.org/integration/mozilla-inbound/rev/da10fe58893d

also, I file bug 742259 for more test.

Comment 10

[(no longer active)]

https://hg.mozilla.org/mozilla-central/rev/da10fe58893d

Comment 11

[Lukas Blakk [:lsblakk] use ?needinfo]

[Triage Comment]
Sounds like a low risk fix - please nominate for Aurora 13 uplift as soon as possible for this regression.

Comment 12

[Jonas Sicking (:sicking) No longer reading bugmail consistently]

Makoto: Does this patch apply to the aurora branch? If so, can you set the approval‑mozilla‑aurora flag for the patch to "?"

Comment 13

[Makoto Kato [:m_kato]]

Attachment: rebase for aurora

[Approval Request Comment]
Regression caused by (bug #):
Bug 590390

User impact if declined:
This is regression by bug 590390.  After landing bug 590390, defalut charset that has no charset in HTTP header isn't UTF-8.  It should keeps UTF-8 as default for compatibility of older version of Firefox.

Testing completed (on m-c, etc.):
Yes.  It has xpcshell test.

Risk to taking this patch (and alternatives if risky): 
Low.  This is regression.

String changes made by this patch:
No.

Comment 14

[Alex Keybl [:akeybl]]

Comment on attachment 613484 [details] [diff] [review]
rebase for aurora

[Triage Comment]
Low risk fix to a regression in FF13. Approved for Aurora.

Comment 15

[Makoto Kato [:m_kato]]

https://hg.mozilla.org/releases/mozilla-aurora/rev/fbd93a3a58f5