Browserscope Results

[rasamassen]

Browserscope's API is admittedly limited (you can finally deleted defunct tests, but you cannot add new tests, reset changed tests, or group tests into categories). As a result, its records for the CSS3Test are not reflecting the actual results (says Chrome18 gets a 63 when my Chrome18 consistently gets a 56).

Based on those limits, one of two things needs to happen:

1. Simple solution is to reset or build a new user test on browserscope that will take in all the new / different data in the CSS3Test as it currently stands. Some form of versioning for the CSS3Test would aid the reality that this test will continually have to be rebuilt from time to time.
2. The more complex solution would be to use some of the backend built for the HTML5Test (with permission, of course), which is built in such a way that it could be relatively easily borrowed, and build a custom results page for the CSS3Test. This would require database access.

[LeaVerou]

This issue has been troubling me for a while too. css3test is hosted on Github pages, and I'd prefer to keep it that way. Which means, server-side coding and databases are out of the question.

I've thought about ditching the old results every time the test is changed, but that would mean deleting tons of valuable data.

Not sure what the solution is. :(

[rasamassen]

Maybe the easiest way would be to only submit overall results and results by CSS3 module. That way, the only time there's a problem with browserscope is when you add or drop a module, not when something in a module changes. So, instead of submitting each individual test, submit the percentage passed in the module. This would relieve a lot of the headache, wouldn't it? Maybe you'd lose the individual results, but it'd give at least some worthwhile data for comparison.

[LeaVerou]

The individual results are VERY important. With proper UI, they could serve as a caniuse for CSS3.

[rasamassen]

Not saying the individual results aren't important. I agree. But while you're using browserscope, you already mentioned that it's a headache without a good solution. So I thought follow browserscope's example where, on for example the ringmark test, it only puts the summary and not each individual test.

I'd love to see a good UI for a caniuse, but you said there's no database option where you're hosting. I'm just suggesting a way to make the browserscope table worthwhile as the individual tests change. Right now it's missing a lot of tests and contains tests for features no longer in existence [like cycle(), min(), and max()]. Combine that with the horizontal table, and it becomes difficult to figure out what data is worthwhile on there, making the whole table more trouble than it's worth. If it contained per-module results, then at least it would much more rarely be out-of-date as the features change and the data would have some level of comparability, like "this browser does decent at flex-box and this browser sucks".

I'm just trying to come up with a solution that makes browserscope manageable and worthwhile until a better way to house the test results is found. I'd like to see how the browsers compare in these features, but I currently find it much easier to open 6 different browsers on my computer and compare them side-by-side than to use the browserscope table.

[LeaVerou]

It doesn't need a database. AFAIK the BrowserScope API can be used client-side. I could build something that gets the test results, compares it with the tests file and displays results only for the currently valid tests.
I still think even the individual test results for other browsers are useful, so that ou know what works and what doesn't.
The problem is that the UI to make them more presentable is missing. The solution to such problems is rarely deleting data, it's usually finding a better way to present them.

[grigs]

Wasn't sure if it made sense to open a new issue for this or not, but Browserscope is currently returning a 500 error for css3test.com: Error in Beacon: KeyTooMany.

Browserscope only allows a max of 200 keys and the current set being passed is over 218 keys.

[LeaVerou]

That’s weird. Is this a new limit? It worked before...

[grigs]

I don't think it is a new limit. My assumption was that you've just added new tests over time. The limit is in the comments of the sample code:
http://www.browserscope.org/user/tests/howto

BTW, thanks for posting the code to css3test. I've been out testing TVs all day and needed to run the test on my own installation so I could have my own browserscope sandbox to record results in.

FWIW, I simply commented out:

//_bTestResults[this.id + ' / ' + feature.replace(/[,=]/g, '')] = Math.round(100 * passed / tests.length)

And worked with the high-level results for the purposes of testing the TVs and keeping the number of keys under 200.

[beatgammit]

This issue has been troubling me for a while too. css3test is hosted on Github pages, and I'd prefer to keep it that way. Which means, server-side coding and databases are out of the question.

What about using iriscouch? It's free for light use, which seems to be a perfect fit for this site. I don't know if they do compactions for you, so you might need to account for that since this will get frequent updates.

[rasamassen]

The browserscope site seems dead. No changes to the code in over a year.

[SebastianZ]

As @rasamassen pointed out, Browser Scope isn't working anymore, so I filed #217 to remove the related code. In #218 we can discuss new ways to save the results. Therefore, I close this issue now.

Sebastian