Start the Debugger on a page with a lot of scripts. There may be some startup slowness populating the list of scripts.

Good example case is twitter.com. Switching scripts with the list is quite slow.

I herd some orion updates will autofix this.

from playing with this in the SPS profiler, the following stand out: _modifyContent > _updatePage > _createLine 20.9% > _createSpan 9.9% … _getLineBoundingClientRect 12.1% >DoReflow 11% _modifyContent and updatePage seem to be pretty heavy hitters, consuming a high percentage of time. They reduce to _createLine and _createSpan being the big culprits in there. I think some targetted profiling around these methods would help. _getLineBoundingClientRect is another big contender causing reflows with every invocation.

(In reply to Rob Campbell [:rc] (:robcee) from comment #3) > from playing with this in the SPS profiler, the following stand out: > > _modifyContent > _updatePage > _createLine 20.9% > _createSpan 9.9% > > … _getLineBoundingClientRect 12.1% >DoReflow 11% > > > _modifyContent and updatePage seem to be pretty heavy hitters, consuming a > high percentage of time. They reduce to _createLine and _createSpan being > the big culprits in there. I think some targetted profiling around these > methods would help. > > _getLineBoundingClientRect is another big contender causing reflows with > every invocation. Awesome profiling is awesome. I guess these are all methods implemented in the orion engine. Maybe showing these results to the orion people would help them?

(In reply to Victor Porof [:vp] from comment #4) > Awesome profiling is awesome. I guess these are all methods implemented in > the orion engine. Maybe showing these results to the orion people would help > them? Indeed. I'm already on it - contacted them. Upstream is aware of the issues, but we haven't come yet to a solution to avoid the numerous calls to "reflow triggers". Thanks Rob for profiling!

Let's check on this again after Orion is updated.

Victor, any update on this?

Nobody is working on this, but I haven't heard any complains after our switch to CodeMirror.

(In reply to Victor Porof [:vporof][:vp] from comment #8) > Nobody is working on this, but I haven't heard any complains after our > switch to CodeMirror. Should be safe to close this bug then. We can always reopen when it becomes a problem again in the future.