Closed
Bug 491454
Opened 15 years ago
Closed 15 years ago
Implement soft and weak references
Categories
(Tamarin Graveyard :: Garbage Collection (mmGC), defect)
Tamarin Graveyard
Garbage Collection (mmGC)
Tracking
(Not tracked)
VERIFIED
INCOMPLETE
Future
People
(Reporter: lhansen, Unassigned)
Details
Attachments
(2 obsolete files)
This includes both C++ APIs (where we already have weak references) and AS3 APIs.
Reporter | ||
Comment 1•15 years ago
|
||
Attachment #375780 -
Flags: review?(treilly)
Updated•15 years ago
|
Attachment #375780 -
Flags: review?(treilly) → review+
Reporter | ||
Comment 2•15 years ago
|
||
Comment on attachment 375780 [details] [diff] [review]
Preliminary cleanup
redux changeset: 5e640ff116ee
Attachment #375780 -
Attachment is obsolete: true
Reporter | ||
Comment 3•15 years ago
|
||
(In reply to comment #2)
> (From update of attachment 375780 [details] [diff] [review])
> redux changeset: 5e640ff116ee
That patch was later reverted because it did not work on all platforms (for reasons not yet known) - the abcasm/coverage/G tests failed (core dump) on win, win64, ppc64, winmo.
Reporter | ||
Comment 4•15 years ago
|
||
Fix confusing and incorrect naming convention in avmcore, and users; fix improper cast to GCObject in hash table code by avoiding the need for a cast.
My confusion in the previous patch was that the AvmCore functions gcObjectToAtom and atomToGCObject and isGCObject don't actually operate on GCObject values at all; they operate on void* and names should reflect that. Fixing the names also exposes the dodgy (but accidentally working) code in the hash tables that needs cleanup: it casts a void* to GCObject but the representation could actually be GCFinalizedObject (which has a vtable and looks different).
Attachment #380437 -
Flags: superreview?(edwsmith)
Attachment #380437 -
Flags: review?(stejohns)
Comment 5•15 years ago
|
||
Comment on attachment 380437 [details] [diff] [review]
Clean up confusing code around weak refs
approved, but could you please use atomPtr(a) instead of (a&~7), and atomKind(a) instead of (a&7)?
Attachment #380437 -
Flags: review?(stejohns) → review+
Updated•15 years ago
|
Attachment #380437 -
Flags: superreview?(edwsmith) → superreview+
Reporter | ||
Comment 6•15 years ago
|
||
Comment on attachment 380437 [details] [diff] [review]
Clean up confusing code around weak refs
redux changeset: 1966:373bf5d9630b
I did clean up the bit operations as requested.
Attachment #380437 -
Attachment is obsolete: true
Reporter | ||
Updated•15 years ago
|
Assignee: lhansen → nobody
Status: ASSIGNED → NEW
Reporter | ||
Updated•15 years ago
|
Target Milestone: --- → Future
Reporter | ||
Comment 7•15 years ago
|
||
Closing this as we have no requirement for those features at this time and no pending patches.
Status: NEW → RESOLVED
Closed: 15 years ago
Resolution: --- → INCOMPLETE
You need to log in
before you can comment on or make changes to this bug.
Description
•