799195 - Old Litmus testing utilities from the RSS tests to port to Moztrap as needed

Status: RESOLVED FIXED

(Mozilla QA Graveyard :: Infrastructure, defect)

Description

[Tanner M. Young [:tmyoung]]

Attachment: Mozilla RSS test files from litmus (some require PHP)

The files in this zip container are from old litmus tests from the RSS tests that I set up three or four years ago.  All of these were hosted at alyoung.com/mozilla/* and were referenced directly in the old tests.  I may have sent these to qa team members before via email before, but in case they are still needed I have posted them so that they can be ported to Moztrap if we still do RSS testing by hand.

I will pull the files off of my server, once this bug is closed.

Comment 1

[Tanner M. Young [:tmyoung]]

One other thing, that I just remembered.  Some of these files run PHP, which allowed us to customize the headers, since many of the tests were header-based, and required the files to be served under special conditions as the RSS handler was and is very picky about protocols and origin issues for security.  If these tests are going to be useful, they must be served in an environment that allows PHP or some comparable utility with control of file headers and mime types.  (Apache .htaccess files with the correct permissions can do the exact same thing, and I am sure there are simple nginx and node.js ways to do this.)

Comment 2

[Henrik Skupin [:whimboo][⌚️UTC+2]]

Thanks Tanner for uploading your testcases! I'm cc'ing Anthony and Juan who could investigate what we want to re-use. Once we figured that out we can get the appropriate tests migrated into the litmus-data repository.

Comment 3

[u279076]

Just looking that the files one by one...

> rss0.91.php: RSS 0.91 feed
> rss1.0.php: RSS 1.0 feed
> rss2.0.php: RSS 2.0 feed
These can easily be converted to xml, I don't see why they need to be PHP.

> rss2.0-1.xml: RSS 2.0 feed
Appears to be a dupe of rss2.0.php, might be unnecessary.

> rss1.0-txt.php
> rss2.0-txt.php
Plain-text versions of the above RSS feeds, not sure why these are needed.

> rss2.0-link.htm
Used to test rss2.0-1.xml

> rss2.0-link.html
> general-rss.htm
Appears to test rss2.0-1.xml as well, not sure why these are needed.

> rss2.0-link.php
Appears to test rss2.0.php

> rss2.0-links.php
Appears to be a dupe of rss2.0-link.php

> bug515549.swf, swf.htm
Tests a bug where plugin content is not shown in the Page Info dialog. We could probably use this.

> xss1.php (XSS test string)
> xss2.php (XSS test page)
Not sure if we need this test at all. It's hard to determine without a referencing regression bug.

Tanner, please comment further to explain these files.

Comment 4

[Tanner M. Young [:tmyoung]]

(In reply to Anthony Hughes, Mozilla QA (:ashughes) from comment #3)
> Just looking that the files one by one...
> 
> > rss0.91.php: RSS 0.91 feed
> > rss1.0.php: RSS 1.0 feed
> > rss2.0.php: RSS 2.0 feed
> These can easily be converted to xml, I don't see why they need to be PHP.

They need to be in php or some equivalent because--as stated in the previous comment--they we are not testing the XML we are testing the behavior of firefox when xml is served with the wrong headers.  There were a number of UX and security problems when the RSS Feed Preview was created where it was detecting RSS feeds where it shouldn't.  For example a text file that contains XML should not automatically become an RSS feed, but there are other cases where it should.  The reason I used PHP was it was the fastest way on my servers to get the files served using headers that would be wrong, intentionally.  So yes, the content on many of these files is duplicate but their header behavior is not.

> > rss2.0-1.xml: RSS 2.0 feed
> Appears to be a dupe of rss2.0.php, might be unnecessary.
This gives the file a header of application/rss+xml instead of just application/xml, which is the default.  In both of the cases--if I recall correctly--Firefox should load feed preview, but I'd have to go find the old litmus test to find out.

> > rss1.0-txt.php
> > rss2.0-txt.php
> Plain-text versions of the above RSS feeds, not sure why these are needed.
These are testing if Firefox will load a feed from a text file, which (and I'd need to look at the tests again to confirm it), shouldn't happen.  Firefox should handle the "content-type: text/plain" header at all costs.  The best way--at the time of serving the files this way was with PHP header commands.  However, you could--if you wanted--have different file extensions for each file and then run an .htaccess file that changed the mime type for each file individually.  (But also there was a time when feed preview may have used file extensions not headers in rare cases, and these tests may also be testing that, I really can't remember.  There are definite origin issues with RSS Feed preview that I remember, like how it handles a file saved to the system, etc.)


> > rss2.0-link.htm
> Used to test rss2.0-1.xml
This is another case of seeing if Firefox behaves the intended way with a marginal case of whether or not to load the file based on extension, headers, or origin.  Over the years RSS Feed-preview used both good and bad assumptions about when something is rss or not.  It would sometimes render to aggressively or too infrequently.  These different tests represent about five to ten different cases of times Firefox should and should not handle malformed/mishandled RSS feeds.  This like on the below I believe tests rss detection.


> > rss2.0-link.html
> > general-rss.htm
> Appears to test rss2.0-1.xml as well, not sure why these are needed.
These test that Firefox correctly detects if a feed is listed on the page.  (The earlier tests focus on when Feed preview should be used or not), these html files focus on if Firefox understands that HTML elements that indicate if an html file is present.  The rss2.0-link.html checks if we can handle both absolute and relative paths--as there was a time where we didn't do it correctly.

> > rss2.0-link.php
> Appears to test rss2.0.php
> 
> > rss2.0-links.php
> Appears to be a dupe of rss2.0-link.php
I believe these represent more of the same (going through the litmus cases will be the easiest for making sense of all of this), but it's been almost four years since I wrote a lot of this, and there may be other reasons for idiosyncrasies and redundancies in this folder, that were requested by developers at the time, because of security bugs and other things that I'm just not remembering at the moment--but everything here, once referenced to the old tests should make decent sense as all of the confusion was created deliberately to test over two dozens bugs with as few files and litmus test cases as possible.


> > bug515549.swf, swf.htm
> Tests a bug where plugin content is not shown in the Page Info dialog. We
> could probably use this.
This is tied to bug 515549, which I never finished.  We were never able to confirm to dveditz and my satisfaction that bug 515549 wouldn't introduce a major security flaw because of flash that we couldn't fully anticipate and test for.  So we had gotten to the point that we could almost test it, and so this was added this to the folder in case we actually ever submitted that bug.  Your welcome to use it, but Page Info wont do anything about this test since the function activates it was never committed to the repository.  If it is, we should by all means add this file to our manual testing.

> > xss1.php (XSS test string)
> > xss2.php (XSS test page)
> Not sure if we need this test at all. It's hard to determine without a
> referencing regression bug.
These are things that were left over from webdev testing projects I did with Stephend.  We needed these hosted separately because they didn't render correctly in litmus.  But now with everything about webdev being automated, I believe these were already incorporated into their respective systems long, long ago.  I didn't realize they got copied into this folder.


> Tanner, please comment further to explain these files.
I hope this makes more sense.  I apologize for the confusion (and abruptness of my replies), there is a lot of convolution that surrounds these files--for good reasons of course--but it's starting to have been long ago enough that I can't remember every single reason for this.  I will look around and see if there is more that I wrote at the time about these files, but this is what I can remember off hand.  Let me know if I can (try) to clarify anything else on this.

Comment 5

[u279076]

Given that we don't really test RSS anymore (it doesn't even have a prominent UI in Firefox anymore) I'm wondering if we need any of this. It would be a shame to lose the files if we did end up needing them though. Perhaps we can stage them somewhere on mozqa.com for safe keeping?

Comment 6

[Henrik Skupin [:whimboo][⌚️UTC+2]]

Sorry for the delay here. There is at least a case for e10s where it could be useful. See bug 1109714. Whatever comes up here I added all the files after an audit to our testcase repository:

https://hg.mozilla.org/qa/testcase-data/rev/5918f8136a8a