XPath vs CSS Performance

[semperos]

What are the current performance differences between using XPath or CSS queries with the various WebDriver drivers for finding elements on the page?

Metrics preferred, from-the-hip analysis from folks who work on the code also welcome.

-Daniel

[Daniel Wagner-Hall]

Short story: Prefer CSS to XPath.

Metrics would be interesting, although would inherently be
micro-benchmarks, and probably quite page-specific. Please do provide
some, I'd be interested to see them!

> --
> You received this message because you are subscribed to the Google Groups
> "Selenium Developers" group.
> To view this discussion on the web visit
> https://groups.google.com/d/msg/selenium-developers/-/RAhIPRqqKNQJ.
> To post to this group, send email to selenium-...@googlegroups.com.
> To unsubscribe from this group, send email to
> selenium-develo...@googlegroups.com.
> For more options, visit this group at
> http://groups.google.com/group/selenium-developers?hl=en.

[semperos]

For cross-browser support, is WebDriver's CSS querying more robust/consistent than XPath? And for that matter, do we have a CSS grammar, even informal, that we can rely on for WebDriver?

This is less a "complaint" about WebDriver documentation, as my own ignorance about competing CSS-style query languages.

[Adam Goucher]

On 12-02-02 2:47 AM, semperos wrote:
> For cross-browser support, is WebDriver's CSS querying more
> robust/consistent than XPath? And for that matter, do we have a CSS
> grammar, even informal, that we can rely on for WebDriver?

http://www.w3.org/TR/css3-selectors/

In RC the grammar was Sizzle (http://sizzlejs.com) which is both a super
and subset of the W3C spec. There are a couple pathological situations
where WebDriver will fall back to Sizzle, but the general rule-of-thumb
is write W3C selectors.

-adam

[Daniel Wagner-Hall]

We expose what the browsers expose, through document.querySelector.
If the browsers don't implement querySelector at all, we patch in CSS3
support, but basically, it's down to what your browser supports.

> --
> You received this message because you are subscribed to the Google Groups
> "Selenium Developers" group.
> To view this discussion on the web visit

> https://groups.google.com/d/msg/selenium-developers/-/qv1GBT8wTLoJ.

[semperos]

One feature present in XPath that I see missing from the CSS spec is the ability to check against the "text" of a given element.

I see an older post here about using the :contains() pseudo function, but trying that in the latest Firefox doesn't work for me. Is there a reliable alternative to XPath's text() function?

[Leo Laskin]

In webdriver there is element.text...  however there is not a specific way to reference text when searching by css.

On Feb 4, 2012 2:37 AM, "semperos" <daniel.l...@gmail.com> wrote:

One feature present in XPath that I see missing from the CSS spec is the ability to check against the "text" of a given element.

I see an older post here about using the :contains() pseudo function, but trying that in the latest Firefox doesn't work for me. Is there a reliable alternative to XPath's text() function?

--

You received this message because you are subscribed to the Google Groups "Selenium Developers" group.

To view this discussion on the web visit https://groups.google.com/d/msg/selenium-developers/-/g-ATPxV0RKQJ.

[Adam Goucher]

:contains() is a sizzle-ism, not standard css which is why its not working.

-adam

> --
> You received this message because you are subscribed to the Google
> Groups "Selenium Developers" group.
> To view this discussion on the web visit

> https://groups.google.com/d/msg/selenium-developers/-/g-ATPxV0RKQJ.

[semperos]

Yeah, that's what I gathered, but was curious if there did exist a "legitimate" alternative.

[Santiago Suarez Ordoñez]

To add some more to this topic, I did some investigation around this last year. Mostly focused in Selenium 1, which relies in external libraries for the queries. I don't think this is a much valuable topic, we WebDriver basically just relies on each browser's implementation (except for IE), so I'm guessing numbers will vary a lot from one to another.

In case you're still interested, here are my conclusions from back then:
http://saucelabs.com/blog/index.php/2011/05/why-css-locators-are-the-way-to-go-vs-xpath/

This is the raw dump of timings:
https://gist.github.com/963821

I think I still have the python script I used to capture all these. I remember it being a bit messy, which prevented me from putting it up in github. I'm definitely willing to share it if you're interested, though.

Best,
Santi

On Sun, Feb 5, 2012 at 5:21 PM, semperos <daniel.l...@gmail.com> wrote:

Yeah, that's what I gathered, but was curious if there did exist a "legitimate" alternative.

--

You received this message because you are subscribed to the Google Groups "Selenium Developers" group.

To view this discussion on the web visit https://groups.google.com/d/msg/selenium-developers/-/VR7C2NEYYMwJ.