Re: Comments on the Mobile Acid test

Hi, Kai.

Thanks a lot for your input.

* Kai Hendry <hendry@iki.fi> wrote:
| I still think it's too difficult without looking at the source to tell
| which tests have failed. Saying I have three red squares doesn't tell
| much. Ok I see you've added numbers, great.

Do you have any suggestions for how this can be further improved? (c:

| Firstly I think we should recognise the Iphone as the baseline. Having
| the iphone fail any tests right now is a little silly IMO...

I'd like to aim higher than the current generation of mobile browsers. The  
iPhone is not on par with desktop browsers yet; neither is Opera. (c:

|  Why? I am sick of XHTML and when I see this test's title somehow
| verifying its existence again on mobile ... XHTML is a huge waste of
| time.

Some UAs claim to support XML, but don't really. I want to test that XML  
is actually parsed as XML with this test. I'll upload a new version of the  
test more suited for this purpose tomorrow.

| CSS3 selectors
|
|  Nice, but since desktop UAs fail this one... a bit of "getting ahead
| of ourselves"?

Possibly. (c:

|  I think we can live without SVG for this test in the light that the
| iphone doesn't support it.

SVG is a requirement from several major operators, and is a standard we  
want browsers, including the next version of the browser on the iPhone, to  
support.

| the data: URI scheme
|
|  Encouraging people to do inline stuff with data:... Is that really a
| good thing?

This test is not all that relevant-- perhaps we should remove it.

| Isn't it better to test for gzip support or something.

That is a good idea. How can we test that?

| XMLHTTPRequest
|
|  Yes! Maybe some more basic JS tests?

Yes, that should be added. Suggestions?

| HTTPS
|
|  GOOD. :) Though which CAs are verified? Is it fair to give Thawte
| all the business as usual? :)

I want to test that the UA supports HTTPS connections, not support for any  
specific CA. So we should use a certificate we should expect any HTTPS  
capable UA to recognize. I guess that leaves us with Thawte and Verisign.

| I think it is very important to test for Internet connection purity.
| Checking that perhaps there isn't some sort of proxy rewriting requests
| type stuff. Know what I mean?

Possibly. Many operator proxies do lots of strange stuff to HTTP requests  
 from mobile browsers. Although I'd prefer if they didn't, I don't think  
that it should be included in this test, as this is beyond the control of  
the UA.

| Maybe some sort of speed test would be good. On desktops it's very
| apparent what the difference is between 56k and broadband. Though less
| so for mobiles.

The performance requirement in the Acid3 test caused a lot of confusion.  
I'd like to avoid that in this test, leaving performance testing to  
performance tests.

| contenteditable
|
|  Not sure about this. Can you offer any more argumentation? I don't
| see the use case.

Many CMSes, blogs, etc. use contenteditable for editing rich text. If we  
want mobile browsers to be able to use the same web applications as  
desktop computers, they need to support this.

| Will try be on a call one day. :) Keep up the great work!

Thanks. (c:

-- 
Wilhelm Joys Andersen
Core QA, Opera Software

Received on Tuesday, 8 April 2008 12:41:25 UTC