503797 - Crash report pages are not well-formed; need escaped ampersand in "Get Help" URL and has an extra </div>

Status: VERIFIED FIXED

(Socorro :: General, task)

Description

[Dave Garrett]

http://crash-stats.mozilla.com/report/index/6fca5dcc-14c0-46a4-b7a0-b630d2090712
http://validator.w3.org/check?uri=http%3A%2F%2Fcrash-stats.mozilla.com%2Freport%2Findex%2F6fca5dcc-14c0-46a4-b7a0-b630d2090712&charset=%28detect+automatically%29&doctype=Inline&group=0

Thanks for the addition of the "Get Help" link on the crash pages. However, its URL has an unescaped ampersand in need of an "&". The hrefs for the cvsblame links are all properly escaped, so it's just this one link that should need changing for this.

The reason I care is that the addition of this is causing an error on parsing in my extension, Crash Report Helper:
https://addons.mozilla.org/en-US/firefox/addon/11217

Of course I could probably find a way to work around this, but it should be escaped so if someone would be so kind as to escape this one little character things should work again.

Comment 1

[Austin King [:ozten]]

We'll try to get this fixed with the next release. Sorry about that.

Comment 2

[Austin King [:ozten]]

Fixed in r1245. Will ship in next release (in next week).

Comment 3

[Dave Garrett]

(In reply to comment #2)
> Fixed in r1245. Will ship in next release (in next week).

Thanks! Wrong revision number though. ;)

Just occurred to me that this would also come up with any signatures with references in them. Whole thing needs escaping if it isn't already.

Comment 4

[Dave Garrett]

Nevermind. Just learned that Bugzilla mis-linkifies it and it's on another SVN. Everything's escaped; thanks again.

Comment 6

[Dave Garrett]

Austin: The validation is also complaining about an tag closing.

Line 2731, Column 14: end tag for element "div" which is not open
        </div> <!-- /PAGE -->
Line 2731, Column 14: XML Parsing Error: Opening and ending tag mismatch: body line 44 and div
        </div> <!-- /PAGE -->
Line 2744, Column 11: XML Parsing Error: Opening and ending tag mismatch: html line 2 and body
    </body>

I'm not sure exactly which one it's complaining about but it should probably get fixed too to get this thing error free.

Comment 7

[Dave Garrett]

Ah, there are two "</div><!-- /details -->" for some reason. I think one needs removal.

Comment 8

[Dave Garrett]

Austin: Yeah, the other error breaks things still. (Do I need to file another bug for this other one? The link in comment 0 listed both errors.)

Comment 9

[Dave Garrett]

Actually, now I think it's complaining about two stray </div> and possibly a </ul>, which wasn't in the validation output with the old version. It may have gotten worse.

(I really want to smack the parser used for XMLHttpRequest.responseXML. Its over-strictness is what's killing Crash Report Helper. Apparently even the most minor of errors causes it to explode.)

Comment 10

[Jesse Ruderman]

You could parse .responseText yourself, but parsing that would be difficult without a patch for bug 102699.

Comment 11

[Austin King [:ozten]]

(In reply to comment #8)
Sorry about that. I'll get this fixed ASAP. No need for another bug.

Comment 12

[Austin King [:ozten]]

(In reply to comment #10)
Not to suggest that you rewrite your code, but FYI all crashes include a link to the JSON formatted gzip compressed file.

http://crash-stats.mozilla.com/dumps/{uuid}.jsonz

So for
http://crash-stats.mozilla.com/report/index/6fca5dcc-14c0-46a4-b7a0-b630d2090712
you can retrieve
http://crash-stats.mozilla.com/dumps/6fca5dcc-14c0-46a4-b7a0-b630d2090712.jsonz

We've all so shipped RSS feeds with this release. Those two formats should be more robust. I think this was a merge issue, and I didn't catch it on stage.

Comment 13

[Dave Garrett]

@Jesse: Yes, I was considering exactly that.

@Austin: Thanks.

As to the JSON, I know about bug 412477. It makes getting at the signature potentially much simpler, but it doesn't really provide a better alternative to the full page for getting the crashing thread modules. The former is what I need for about:crashes signature listing and the later is what I need to analyze things and identify the crashing plugin or extension (if possible). The current JSON implementation just has the dump in a single text blob so I'd have to parse it myself to get at that information. Of course I could do that but I'd rather take the information from the server which has already been reliably parsed. (is there a doc somewhere on this dump format?) For the moment, things should work fine so long as the crash page is error-free, but I'd I might like to get some additional JSON or XML access at some point in the future.

Comment 14

[Austin King [:ozten]]

The facepalm cycle has completed.

Comment 15

[Jesse Ruderman]

Crash Reporter Helper works now :)  Tested in Firefox 3.5.

Comment 16

[(not currently active) Ted Mielczarek]

Dave: there aren't any docs on the raw dump format, but it's not terribly complicated. Feel free to ping me offline about it if you're interested.

Comment 17

[Dave Garrett]

"This document was successfully checked as XHTML 1.0 Strict!" :)

They don't call it strict for nothing. Thank you very much for the fixes. Everything works perfectly again.

Ted: Thanks for the offer. I'll take a look at it in a bit. Will probably be simpler overall than getting more JSON for all the info, seeing as I only need one portion.