Handle relative paths when sanitizing URLs

[utkarsharma2]

In the initial PR(#41665) we didn't handle the relative path in URL which led to issue(#41977). This PR aims at handling the relative path case when sanitizing URLs

The issue is fixed with current change:

[utkarsharma2]

Not entirely sure where to add test cases, if someone can point me to it?

[potiuk]

Not sure if we have some tests for this one ? @pierrejeambrun@bbovenzi ?

[potiuk]

LGTM now. I never imagined such a simple thing in javascript needs to be done in such way (but it's much more reasonable than the initial way).

[eladkal]

@utkarsharma2 This will target 2.10.2?

[potiuk]

I marked it so - but this was I think what we agreed on, that it is not going to cancel rc1

[utkarsharma2]

Thanks @potiuk, yes this belongs to 2.10.2.

[mfatemipour]

Hi, Could we have this in 2.10 ?

[potiuk]

It is already released in 2.10.2 - have you checked it @mfatemipour ? Can you please verify it in release notes and upgrade (2.10.4 is now available)/ test locally and confirm that it works? If not, open a new issue describing the problem you have.

Thanks for your help in advance.

[gpadavala]

I am on 2.10.3 and this is still happening, on checking it looks like the fix is not in the releases, the code is visible in the main branch but not in the tags 2.10.4 / 2.10.3

[potiuk]

I am on 2.10.3 and this is still happening, on checking it looks like the fix is not in the releases, the code is visible in the main branch but not in the tags 2.10.4 / 2.10.3

Interesting: @utkarsharma2 -> it's marked with 2.10.2 milestone, but yes it seems it's not cherry-picked there.

[potiuk]

I changed milestone to 2.10.5 and will cherry-pick it for 2.10.5

[potiuk]

But it would be great to find what happened :)

[utkarsharma2]

But it would be great to find what happened :)

I think I missed backporting it, somehow. I will investigate more. :/

Thanks for the backporting it.

[potiuk]

I think I missed backporting it, somehow. I will investigate more. :/

Yeah. Could be also some tooling issue of ours :(. I think we should do some extra scrutiny for 2.10.5 (with @kaxil on board as well). This is actually the problem with release management tools, that when they work, we start trusting them and stop checking and validating things, bue it's pretty natural that some refactorings introduce some new cases - including edge cases, so once in a while we should pretty manually verify all the steps of release process, otherwise we might miss something.

Having this check for 2.10.5 might be a good idea to do it in preparation for 3.0.0 (though there we will have to do extra-extra-scrutiny as there will be MANY NEW THINGS). At least if we find and fix some things in 2.10.5 we will have less to do when 3.0.0 comes.