Hi all, currently, we are using phantomjs browser for running Node.js tests [1]. However, we are experiencing some issues which seem to be caused by phantomjs, so we are considering to change the default browser from phantomjs to headless chrome. Do you have some objections why not to do so? [1] https://github.com/keycloak/keycloak-nodejs-connect/blob/master/test/util... _______________________________________________ keycloak-dev mailing list keycloak-dev(a)lists.jboss.org https://lists.jboss.org/mailman/listinfo/keycloak-dev

Hi Michal, I'm not against changing the browser used for testing, as long as we keep Travis happy, like Vašek mentioned. Before we make this decision, I suggest to first isolate the tests and give it a try with some headless browser (maybe chrome-headless). After that, we can decide whether or not we should drop phantomjs. On 2019-05-22, Michal Hajas wrote: > Hi all, > > currently, we are using phantomjs browser for running Node.js tests [1]. > However, we are experiencing some issues which seem to be caused by > phantomjs, so we are considering to change the default browser from > phantomjs to headless chrome. > > Do you have some objections why not to do so? > > [1] > https://github.com/keycloak/keycloak-nodejs-connect/blob/master/test/util... > _______________________________________________ > keycloak-dev mailing list > keycloak-dev(a)lists.jboss.org > https://lists.jboss.org/mailman/listinfo/keycloak-dev -- abstractj

+1 as long as chrome-headless works everywhere I'd rather see that as the default and stop using phantom-js and htmlunit (or whatever it's called). From looking at this a looooong time ago I seem to remember chrome-headless was faster than everything else. On Thu, 23 May 2019 at 08:51, Michal Hajas <mhajas(a)redhat.com&gt; wrote: > I created an issue [1] and I will start to work on making it work on > Travis. > > [1] https://issues.jboss.org/browse/KEYCLOAK-10359 > > On Wed, May 22, 2019 at 7:46 PM Bruno Oliveira <bruno(a)abstractj.org&gt; > wrote: > > > Hi Michal, I'm not against changing the browser used for testing, as > > long as we keep Travis happy, like Vašek mentioned. > > > > Before we make this decision, I suggest to first isolate the tests and > > give it a try with some headless browser (maybe chrome-headless). After > > that, we can decide whether or not we should drop phantomjs. > > > > On 2019-05-22, Michal Hajas wrote: > > > Hi all, > > > > > > currently, we are using phantomjs browser for running Node.js tests > [1]. > > > However, we are experiencing some issues which seem to be caused by > > > phantomjs, so we are considering to change the default browser from > > > phantomjs to headless chrome. > > > > > > Do you have some objections why not to do so? > > > > > > [1] > > > > > > https://github.com/keycloak/keycloak-nodejs-connect/blob/master/test/util... > > > _______________________________________________ > > > keycloak-dev mailing list > > > keycloak-dev(a)lists.jboss.org > > > https://lists.jboss.org/mailman/listinfo/keycloak-dev > > > > -- > > > > abstractj > > > _______________________________________________ > keycloak-dev mailing list > keycloak-dev(a)lists.jboss.org > https://lists.jboss.org/mailman/listinfo/keycloak-dev

On 23/05/2019 09:48, Stian Thorgersen wrote: > +1 as long as chrome-headless works everywhere I'd rather see that as the > default and stop using phantom-js and htmlunit (or whatever it's called). > From looking at this a looooong time ago I seem to remember chrome-headless > was faster than everything else. I think this is not the case when comparing to HtmlUnit. HtmlUnit was far faster than other browsers due the fact that it's just wrapper over apache HTTP client. All the other web driver implementations are subclass of RemoteWebDriver and they use remote selenium server as another layer when communicating with the browser. This adds the most of the overhead as there is much more network communications here and there. By using the HtmlUnit instead of phantomJS as default browser, the testsuite was made 2 times faster. AFAIR there wasn't any subtle difference between PhantomJS and other "real" browser WebDriver implementations as all just use selenium server. At least that was the case few years ago when I was looking at it. The thing is, that HtmlUnit has quite bad javascript support, which can be an issue once we have new Account console and maybe some more javascript support in our authenticators. Marek > > On Thu, 23 May 2019 at 08:51, Michal Hajas <mhajas(a)redhat.com&gt; wrote: > >> I created an issue [1] and I will start to work on making it work on >> Travis. >> >> [1] https://issues.jboss.org/browse/KEYCLOAK-10359 >> >> On Wed, May 22, 2019 at 7:46 PM Bruno Oliveira <bruno(a)abstractj.org&gt; >> wrote: >> >>> Hi Michal, I'm not against changing the browser used for testing, as >>> long as we keep Travis happy, like Vašek mentioned. >>> >>> Before we make this decision, I suggest to first isolate the tests and >>> give it a try with some headless browser (maybe chrome-headless). After >>> that, we can decide whether or not we should drop phantomjs. >>> >>> On 2019-05-22, Michal Hajas wrote: >>>> Hi all, >>>> >>>> currently, we are using phantomjs browser for running Node.js tests >> [1]. >>>> However, we are experiencing some issues which seem to be caused by >>>> phantomjs, so we are considering to change the default browser from >>>> phantomjs to headless chrome. >>>> >>>> Do you have some objections why not to do so? >>>> >>>> [1] >>>> >> https://github.com/keycloak/keycloak-nodejs-connect/blob/master/test/util... >>>> _______________________________________________ >>>> keycloak-dev mailing list >>>> keycloak-dev(a)lists.jboss.org >>>> https://lists.jboss.org/mailman/listinfo/keycloak-dev >>> -- >>> >>> abstractj >>> >> _______________________________________________ >> keycloak-dev mailing list >> keycloak-dev(a)lists.jboss.org >> https://lists.jboss.org/mailman/listinfo/keycloak-dev > _______________________________________________ > keycloak-dev mailing list > keycloak-dev(a)lists.jboss.org > https://lists.jboss.org/mailman/listinfo/keycloak-dev

Hello Michal, you probably already see it but there is a section regarding chrome in the travis doc https://docs.travis-ci.com/user/gui-and-headless-browsers/#using-the-chro... If you are recording the test suite you will need a display server (xvfb works great, also in the same doc) Best On Tue, May 28, 2019 at 9:15 AM Stian Thorgersen <sthorger(a)redhat.com&gt; wrote: > The JS engine used in HtmlUnit is not very fast and also has other issues. > > Perhaps the case is we should default to HtmlUnit when there is no JS > involved (login flows), and use Chrome Headless when there is JS involved > (account, admin, keycloak.js and node.js). > > On Mon, 27 May 2019 at 21:50, Marek Posolda <mposolda(a)redhat.com&gt; wrote: > > > On 23/05/2019 09:48, Stian Thorgersen wrote: > > > +1 as long as chrome-headless works everywhere I'd rather see that as > the > > > default and stop using phantom-js and htmlunit (or whatever it's > called). > > > From looking at this a looooong time ago I seem to remember > > chrome-headless > > > was faster than everything else. > > > > I think this is not the case when comparing to HtmlUnit. > > > > HtmlUnit was far faster than other browsers due the fact that it's just > > wrapper over apache HTTP client. All the other web driver > > implementations are subclass of RemoteWebDriver and they use remote > > selenium server as another layer when communicating with the browser. > > This adds the most of the overhead as there is much more network > > communications here and there. > > > > By using the HtmlUnit instead of phantomJS as default browser, the > > testsuite was made 2 times faster. AFAIR there wasn't any subtle > > difference between PhantomJS and other "real" browser WebDriver > > implementations as all just use selenium server. At least that was the > > case few years ago when I was looking at it. > > > > The thing is, that HtmlUnit has quite bad javascript support, which can > > be an issue once we have new Account console and maybe some more > > javascript support in our authenticators. > > > > Marek > > > > > > > > On Thu, 23 May 2019 at 08:51, Michal Hajas <mhajas(a)redhat.com&gt; wrote: > > > > > >> I created an issue [1] and I will start to work on making it work on > > >> Travis. > > >> > > >> [1] https://issues.jboss.org/browse/KEYCLOAK-10359 > > >> > > >> On Wed, May 22, 2019 at 7:46 PM Bruno Oliveira <bruno(a)abstractj.org > > > >> wrote: > > >> > > >>> Hi Michal, I'm not against changing the browser used for testing, as > > >>> long as we keep Travis happy, like Vašek mentioned. > > >>> > > >>> Before we make this decision, I suggest to first isolate the tests > and > > >>> give it a try with some headless browser (maybe chrome-headless). > After > > >>> that, we can decide whether or not we should drop phantomjs. > > >>> > > >>> On 2019-05-22, Michal Hajas wrote: > > >>>> Hi all, > > >>>> > > >>>> currently, we are using phantomjs browser for running Node.js tests > > >> [1]. > > >>>> However, we are experiencing some issues which seem to be caused by > > >>>> phantomjs, so we are considering to change the default browser from > > >>>> phantomjs to headless chrome. > > >>>> > > >>>> Do you have some objections why not to do so? > > >>>> > > >>>> [1] > > >>>> > > >> > > > https://github.com/keycloak/keycloak-nodejs-connect/blob/master/test/util... > > >>>> _______________________________________________ > > >>>> keycloak-dev mailing list > > >>>> keycloak-dev(a)lists.jboss.org > > >>>> https://lists.jboss.org/mailman/listinfo/keycloak-dev > > >>> -- > > >>> > > >>> abstractj > > >>> > > >> _______________________________________________ > > >> keycloak-dev mailing list > > >> keycloak-dev(a)lists.jboss.org > > >> https://lists.jboss.org/mailman/listinfo/keycloak-dev > > > _______________________________________________ > > > keycloak-dev mailing list > > > keycloak-dev(a)lists.jboss.org > > > https://lists.jboss.org/mailman/listinfo/keycloak-dev > > > > > > > _______________________________________________ > keycloak-dev mailing list > keycloak-dev(a)lists.jboss.org > https://lists.jboss.org/mailman/listinfo/keycloak-dev -- Guillaume Vincent Senior Software Engineer Red Hat <https://www.redhat.com> <https://www.redhat.com> _______________________________________________ keycloak-dev mailing list keycloak-dev(a)lists.jboss.org https://lists.jboss.org/mailman/listinfo/keycloak-dev