Closed Bug 84977 Opened 24 years ago Closed 24 years ago

News can't open one of my newsgroups (INVALID_POINTER error)

Categories

(SeaMonkey :: MailNews: Message Display, defect)

x86
Windows 98
defect
Not set
blocker

Tracking

(Not tracked)

VERIFIED FIXED
mozilla0.9.3

People

(Reporter: SkewerMZ, Assigned: Bienvenu)

Details

(Whiteboard: [nsbeta1+])

Procedure: Difficult to reproduce procedure, this somehow occurred after weeks of use. Tried to load newsgroup alt.html, have two news accounts. Expected: Newsgroup opens. Actual: Newsgroup doesn't open. This pops into my JS console: Error: uncaught exception: [Exception... "Component returned failure code: 0x80004003 (NS_ERROR_INVALID_POINTER) [nsIMsgFolder.getMsgDatabase]" nsresult: "0x80004003 (NS_ERROR_INVALID_POINTER)" location: "JS frame :: chrome://messenger/content/commandglue.js :: ChangeFolderByDOMNode :: line 118" data: no] Source File: Line: 7 Build: 2001060804 Win98. This has happened in previous builds.
Keywords: nsbeta1, nsCatFood
This is a duplicate. (Bienvenu, we chatted in that bug, remember?)
looks like the db might have been corrupted...do you ever run two versions of the client against the same profile?
I think the database corruption may have something to do with my frequent encounters of bug 83486.
No, I don't think so. bug 83486 has to do with partial, incomplete messages getting into the memory cache and has nothing to do with the database.
Temporary, horrible workaround: Unsubscribe and resubscribe to the faulty newsgroup.
QA Contact: esther → stephend
I'll fix the handling of the corrupted database, though I don't know the cause of the corruption (other than running two clients against the same profile).
Assignee: sspitzer → bienvenu
As I reported in bug 85920, I had a similar problem. However, the suggested workaround of unsubscribing was not useable. Merely selecting the problem group was enough to crash Mozilla. Fortunately, stephend suggested deleting the msf files, and that, along with deleting the rc file, seems to have cured the problem. As he suspected, I've upgraded Mozilla from version to version without making any change to the profile.
you can unsubscribe without selecting the newsgroup by using the subscribe UI, which is what Stephen meant, I believe. Unsubscribing merely deletes the .msf file
I should note that I've seen the crash myself, but only at work (after one day running Moz with Netscape 6.1 concurrently, and they're sharing the same profile). I believe it's safe to say that since I haven't seen this crash on my home machine which uses only the latest Moz trunk builds, that this is a cause of the bug.
Perhaps it is a form of data base corruption. But my environment does not use any browser except Mozilla. I started with Msomething, then went through 0.7 to 0.9.1 with a few non-release builds scattered in there. Would there be value in looking at my old msf file? I've saved a copy.
moving to 0.9.3
Whiteboard: [nsbeta1+]
Target Milestone: --- → mozilla0.9.3
I checked in code to the trunk so we shouldn't get that js assertion anymore, and if we have a corrupted .msf file, we should just redownload all the headers again. If you have a test case, please try it out and let me know if it works. Thanks.
marking fixed - please reopen if you see this again (only fixed on the trunk, not the branch)
Status: NEW → RESOLVED
Closed: 24 years ago
Resolution: --- → FIXED
Skewer, do you see this at all anymore? I don't see it with my trunk build (2001-06-29-04) on Windows 2000.
I will reopen this if I get any more corruption problems. Currently my profile is stored in a Windows folder away from Netscape (in fact NS6.01 is completely gone) so if anything happens Netscape won't be the cause.
I haven't seen this at home with an old profile, nor with a newly created profile (and I have Netscape 6.1 PR1 running alongside moz). Verifying with build 2001-07-10-09. We should refile this if we see it again, but like I said, I haven't.
Status: RESOLVED → VERIFIED
Product: Browser → Seamonkey
You need to log in before you can comment on or make changes to this bug.