Injected DOM wrappers were a nice idea in theory but everyone just passes
wrappers around document.createElement/document.createTextNode. That
includes HtmlReporter's unit tests and karma-jasmine-html-reporter, the
only known 5.x-compatible library that constructs an HtmlReporter.
This change broke spec filtering in Karma by changing the format of the
`spec` query parameter. Although karma-jasmine-html-reporter uses
jasmine-core's HtmlSpecFilter, karm-jasmine provides its own spec filter
that interprets the query parameters itself.
This feature may be reintroduced in 6.0 as a breaking change.
This reverts commit 8309416cb2.
The previous commit left one code path un-converted to globalThis. That
exposed a bug in Karma: Karma loads jasmine-core via <script type="module">
even though it's not an ES module. In ES modules, the value of `this`
outside of a method is undefined rather than window. The Angular dev tools
try to work around that bug by monkey-patching window to look like GJS's
window object, which worked until the GJS hack was removed in the previous
commit.
This slightly simplifies the init code and hardens Jasmine against broken
bundlers that assume everything can be wrapped in a "use strict" context.
This removes a workaround for incompatible `this` behavior in GJS. GJS was
never a supported envronment, but in any case the change is unlikely to
cause problems since GJS ha supported globalThis since 2020.
This reverts commit e2a7740322.
structuredClone errors in these browsers were a symptom of inavertently
including Error objects in reporter events. In newer browsers,
structuredClone can copy those objects, but it's lossy: if an instance of
an Error subclass is cloned, the result is an instance of Error.
With that fixed, Jasmine is compatible with Safari 16 and FF 102. At least
for now. And keeping them around may provide a way to detect similar bugs.
This reverts commit 6da88ec19eea2780c030dc95fcc55d059fed69e5.
Removing existing handlers turns out to be load-bearing for Jasmine's
parallel mode. ParallelWorker (in the jasmine package) installs a pair
of handlers before booting core so that it can catch late async errors
that happen after one spec file has finished executing and before the
next starts. If those aren't uninstalled, errors that get routed through
jasmine-core's normal error handling mechanism will also be reported
via ParallelWorker's handlers. It might be possible for ParallelWorker
to uninstall and install its handlers at the right time, but it's
likely that there would be gaps in between when core uninstalls its
handlers and when ParallelWorker installs. And in any case, the
old behavior of GlobalErrors was a perfect match for what ParallelWorker
needs, so let's keep it.
