- From: <bugzilla@jessica.w3.org>
- Date: Wed, 02 Apr 2014 21:01:28 +0000
- To: public-webapps-bugzilla@w3.org
https://www.w3.org/Bugs/Public/show_bug.cgi?id=24576 --- Comment #24 from Arun <arun@mozilla.com> --- (In reply to Anne from comment #22) > So we want something that works for bug 25081 comment 6. > > If a blob that is closed simply loses the reference to its underlying source > and returns an empty byte sequence, that does not seem great for FormData as > it would be somewhat racy. However, returning the empty byte sequence if its > closed seems good. > > Maybe blobs should have a keepalive list of objects. Then at the point where > a FormData object (FD) is serialized, FD adds itself to the keepalive list. > Then when FD starts reading from the blob it passes itself as a parameter to > read and is allowed to bypass the closed check and reach the source because > it is on the keepalive list. Then once the read completes FD removes itself > from the list and if the blob is closed and has an empty keepalive list it > can GC. > > Does that model make sense? After today's discussion in IRC, this model makes sense. It would be a keepalive list of objects using a given Blob. FileReader should add itself to the keepalive list of objects using a given Blob, too, and then remove itself. -- You are receiving this mail because: You are the QA Contact for the bug.
Received on Wednesday, 2 April 2014 21:01:29 UTC