fix(compiler): handle null window.location.origin

[brunor97]

when using a data URI or file URI (#2582), or when using iframe's srcdoc, the value for window.location.origin can be null. In these particular cases, passing null to new URL(url, null) will throw the following error: Failed to construct 'URL': Invalid base URL

[00salmon]

+1, currently blocked by this when using components in an iframe

[brunor97]

+1, currently blocked by this when using components in an iframe

As a hotfix, I'm using patch-package to apply this fix on my project's side, while we wait for this PR to be merged and published

Not a long term solution but hopefully this can unblock you 😃

[00salmon]

+1, currently blocked by this when using components in an iframe

As a hotfix, I'm using patch-package to apply this fix on my project's side, while we wait for this PR to be merged and published

Not a long term solution but hopefully this can unblock you 😃

that's more convenient than editing the file in dist directly before each publish, thank you!

[brunor97]

+1, currently blocked by this when using components in an iframe

As a hotfix, I'm using patch-package to apply this fix on my project's side, while we wait for this PR to be merged and published
Not a long term solution but hopefully this can unblock you 😃

that's more convenient than editing the file in dist directly before each publish, thank you!

No problem! Glad I could be of help 😄

[Sanderand]

+1, we make use of the patch-package solution for this issue. for @Stencil@[email protected] the patch file is 8.5mb and needs to be checked into our repo. 🥲

will this PR eventually be considered?

[loganvolkers]

Bump @rwaskiewicz

If this needs some changes to get it over the line, please let me know. We have dozens of stencil projects that all need this fix manually applied so they are compatible with sandboxed iframes and srcdoc iframes

[loganvolkers]

e.g. I'm happy to add some test cases to this PR if that will help it get across the line

[jan-paulus]

Any update on this?

[Sanderand]

Any update on this?

[leonheess]

@rwaskiewicz, please merge these changes or outline any requirements that need to be fulfilled

[jan-paulus]

@rwaskiewicz Could you take a look?

We've used patch-package to workaround this issue for more than a year now.

[rwaskiewicz]

LGTM - I have one stylistic nitpick that I think can be applied in GH

[rwaskiewicz]

I'm going to be merging this as an admin - the failing CI step is a known issue when running against forks. The step in question is our tech debt burndown chart, which I'm able to see the output of (no new unused functions or strict null checks violations are added in this PR)