OS.File loads scripts in a way that means they stick around in memory as strings
Categories
(Toolkit Graveyard :: OS.File, defect, P3)
Tracking
(firefox67 affected)
Tracking | Status | |
---|---|---|
firefox67 | --- | affected |
People
(Reporter: Gijs, Unassigned)
References
(Blocks 1 open bug)
Details
(Whiteboard: [fxperf:p3])
about:memory shows 5 strings with total length around 90-100k and thus taking around 200k of memory (yay ucs2 + js overhead, I guess).
This is in addition to the actual interpreted/compiled representations.
We should try to avoid this. It's probably a consequence of the use of require.js - perhaps now we can "just" use es6 modules or some other, better mechanism.
Reporter | ||
Updated•6 years ago
|
It might also be the DOM holding one copy of the string and JS holding another copy.
Reporter | ||
Comment 2•6 years ago
|
||
(In reply to David Teller [:Yoric] (please use "needinfo") from comment #1)
It might also be the DOM holding one copy of the string and JS holding another copy.
Sorry, what exactly do you mean? Workers don't have a DOM, right?
They don't have access to the document but they have access to some DOM APIs.
Here, it's XHR used in a worker.
Comment 4•6 years ago
|
||
Marking as P3 because it seems only related to memory usage. However, if someone proposes a solution that would also improve the load times for OS.File, we would probably bump the priority.
I thought that memory usage was back to being a priority?
Reporter | ||
Comment 6•6 years ago
|
||
(In reply to David Teller [:Yoric] (please use "needinfo") from comment #5)
I thought that memory usage was back to being a priority?
We're focused on startup time and mainthread IO for the moment.
Updated•2 years ago
|
Updated•2 years ago
|
Comment 8•2 years ago
|
||
OS.File is being removed.
Updated•2 years ago
|
Description
•