firstname.lastname@example.org sent me an email regarding the thread safety of nsIFileStream. Currently this class is not thread safe and is asserting on addref/release. I do not think that the right thing to do is merely remove the asserts, as they correctly point out a serious problem. mscott, maybe you want to look at this. From Tony's email: ---- Subject: Threadsafe nsIFileStream? Date: Sat, 11 Mar 2000 09:34:46 -0800 From: "Tony Robinson" <email@example.com> To: <firstname.lastname@example.org> Somehow, in mailnews after importing mail, an nsMsgFolder releases a nsIFileSpec when mozilla quits that winds up releasing a stream and I hit those great new threadsafe asserts. Should I try and track down exactly why this happens (could be lots of work!) or is it OK to fix up the AddRef/Release in nsIFileStream.cpp? ---
OK. If that class's AddRef/Release should not be made threadsafe then I need to find out exactly why I'm seeing this.
Assignee: mscott → tonyr
Duh... I was relying on nsIFileSpec to shut down any open streams when it was released but of course, the nsMsgFolder owned the file spec so it wouldn't get cleaned up till much much later. A simple call to CloseStream from the import solved this problem.
Status: NEW → RESOLVED
Last Resolved: 18 years ago
Resolution: --- → FIXED
I checked this out and see the calls to CloseStream(). - rhp
Status: RESOLVED → VERIFIED
You need to log in before you can comment on or make changes to this bug.