SEF on: Switcher broken => Switching to another home page impossible

As I have merged #5135, this will anyway need update

Its updated. I will have a look at the rest later.

Anyway, this would not be B/C at all. Also we would have other places in code where the prefix is used and not the language tag. Not for this series imho.

Right now, it depends on your definition of BC. Because it still parses the old URLs and it also still processes the old URLs, since there are to many extensions out there, that are using the current Joomla style of generating these URLs. If you can live with the change for the current non-SEF URLs, then this is still BC.

As long as we propose in Global Configuration SEF as an option, this would not be B/C.

As I said earlier: The old URLs still work. We would only be introducing new URLs for the future.

If existing URLs and features continue to work as they are today, there's no need for a config option IMO.

With these latest changes, this would be B/C. These changes require #5264 to be applied, too.

Besides these changes, I just wanted to make you aware of it that we are not using the language parameter of any article in our system right now in the routing... Actually, I searched the whole system and out of 35 files, we are only using the language parameter in 3 of those and there are maybe 10 files where we are not even using the category ID, thus making the resulting URL complete garbage. Will write a PR for that...

Test: not good here
I first applied #5264

SEF on:
Switching between homes does not work (When url lang code is set to off for default site lang)
Switching between associations => 404
SEF off:
we don't even get the site url:
http://localhost:8888/?lang=en-GB

Please check this again. I invested several hours into this now to work mostly through it and this should be good now. I would call this merge-ready right now. Again, as I wrote earlier, this requires #5264 to work.

@weeblr-dev Please have a look over this, too. Would this still work properly with your system?

@zero-24 Please do CS stuff afterwards. As long as Travis is okay with the current code, I'd rather not change it again and enforce another round of tests just because of that.

ok. I will add a PR after the merge here.

Regarding the search bug: Fixed it.

Thanks looks fixed with: Hackwar@b8127ed

Regarding the bug: Again the old implementation gets us here. In the old implementation, the first part of the URL was always parsed as a language and in the actual parsing routine, nothing could majorly be broken. Now however, I was first checking if we were called via a POST request. If that is the case, the rest would not be checked. But a POST URL also contains a SEF part and thus we first need to discover that part and then check if we are called via POST.

Please note that the last commit is really simply just a codestyle commit and does not require new tests.

Now that we have 2 tests by @richard67 and @zero-24, could this be set to RTC? 😉

Hmm, from my point of view it is not only code style to remove this unused if, it is a refactoring. I trust you, but I know by experience from myself that errors can happen even with simple things. So I will retest anyway and let you know the result.
Regarding RTC: I also want to have this into 3.4, but as we have seen it is such a interdependent thing and - even after your large cleanup, millions of thanks for this - still some kind of complex, so I would feel safer if @infograf768 had a successful test, too. Just my opinion.

@richard67 The code was the following:

$found = false;
if (!$found) {
//do stuff
}

That if will always execute, come hail or storm, so it was simply a useless line of code. 😉

@Hackwar Yes, I just have checked it with Beyond Compare, which is a bit more useful than the stupid GitHub diff, and it looks safe.

@Hackwar You know the code change above was clear to me also by the GitHub diff, but it was not so obvious if not accidently anything else had been changed within the if beside indention, e.g. because you have hit the del key with your head when falling asleep after this much work in this PR ;-) Maybe call me paranoid ;-)

No problem.

@test successful - on my bilingual site everything still works after the latest change. Didn't detect any issue when switching languages with SEF enabled/disabled.

@test tested it on a site with three different languages. Couldn't find any issues while looking around and testing the language stuff. Looks good to me.

@test

looks fine now here with my usual tests.

Based on several testing i think we can move it to RTC now. Thanks to all for the hard work on this!

Thanks. Merged with a9c7eb2

Thank you