It seems the discussion is spread between this thread and the JIRA https://issues.apache.org/jira/browse/WICKET-6544?focusedCommentId=16423835&page=com.atlassian.jira.plugin.system.issuetabpanels%3Acomment-tabpanel#comment-16423835
As far as I can see we don't have consensus if this feature should 1) remain as is (drop PR) 2) be improved (merge PR and/or enhance detection) 3) browser detection should be dropped? I would vote for option 2+ :) On Mon, Apr 2, 2018 at 5:11 AM, Martin Grigorov <mgrigo...@apache.org> wrote: > On Thu, Mar 29, 2018 at 1:31 AM, Korbinian Bachl < > korbinian.ba...@whiskyworld.de> wrote: > > > > > > > ----- Ursprüngliche Mail ----- > > >> even in 2009 it was considered bad: https://www.sitepoint.com/why- > > >> browser-sniffing-stinks/ > > >> and in case that is not enough, read what the guy that invented > > modernizr > > >> has to say: > > >> http://farukat.es/journal/2011/02/499-lest-we-forget-or- > > >> how-i-learned-whats-so-bad-about-browser-sniffing/ > > >> > > >> > > > I do not trust anyone who says "don't do it this way" but doesn't say > how > > > to do it! > > > > > > There are several of "if (isBrowserX()) {...} else {...}" in Wicket JS > > code > > > and they served well for the last decade. > > > Since there are several other *Java* libraries for user agent detection > > > this means that someone still finds them useful despite what other > people > > > claim. > > > > unreliable things wont get reliably by pointing into the past and then > > telling that your fater did it the same way.... > > > > nowadays you would use feature detection, see: > > > > https://developer.mozilla.org/en-US/docs/Learn/Tools_and_ > > testing/Cross_browser_testing/Feature_detection > > > Korbinian, > > The PR by Maxim is about the User-Agent detection at the *server* side, > i.e. in the *Java* code. It reads the request header and tells you what the > browser is. > The JS feature detection is only client side. You will need Ajax behaviors > to send the ourcome to the server to be able to use it there. Wicket does > this with (Web)ClientInfo related classes. > > I'll be VERY glad to see your PR that uses modern ways to redo the current > checks in wicket-ajax.js or in the server code, e.g. Wicket Bootstrap uses > this information to decide whether to render respond.js! > Until then please do not make such bold statements. It is easy to read an > article and say "this is the [new] silver bullet". Until you get your hands > dirty you never know what kind of problems you will face! > > > > > > > > > > > > > > >> btw: > > >> https://github.com/HaraldWalker/user-agent-utils -> this is EOL, > guess > > >> why... > > >> https://github.com/pieroxy/java-user-agent-detection/releases -> last > > >> release from september 2017... > > >> > > >> > > > Sep 2017 is like yesterday > > > > (all only MAJOR releases!) > > > > 28. September 2017 - Firefox 56 > > 14. November 2017 - Firefox 57 Quantum > > 23. Januar 2018 - Firefox 58 > > 13. März 2018 - Firefox 59 > > > > 2017-09-05 - Chrome 61.0.3163 > > 2017-10-17 - Chrome 62.0.3202 > > 2017-12-05 - Chrome 63.0.3239 > > 2018-01-23 - Chrome 64.0.3282 > > 2018-03-06 - Chrome 65.0.3325 > > > > and this is just 2 desktop ones! I dont want to talk about the loads of > > updates my android device got in that time (firefox mobile, chrome and > > samsung internet!) - oh, and btw: they still lie about the user agent all > > time.... dont get me wrong, but sep 17 is freaking old in case you need > to > > reliably detect the browser! > > > > Yes, and all of them are properly parsed by the same code that has been > used in the last decade! > The browser vendors have no reason to change their syntax of user agent. > Believe me they do know that this piece of information *is being* used in > the wild! > -- WBR Maxim aka solomax
