If you ask Mozilla to load a webpage whose server is glitching, and the HTTP dialogue aborts with a network-level error such as "Connection reset", Mozilla will simply print "document: done" on the status bar and remain on the previous page. This is extremely unhelpful. All errors reported by the operating system should result in an error message displayed to the user, and the error message should always contain the printable name of the operating system error (from strerror() or equivalent). Where possible, Mozilla should load a previously-cached copy of the page when the server refuses to communicate with it (Netscape 4.0 would do this). Incidentally, "displayed to the user" does _not_ mean printed to mozilla's standard error with the debugging messages - most people redirect that to the bit bucket. Put up a dialog box, or print it in the status bar.

I also occasionally see this on Win98.

Probably a duplicate as well of depenency for bug 19073.

Marking NEW dont believe this is a dupe.

Bugs targeted at mozilla1.0 without the mozilla1.0 keyword moved to mozilla1.0.1 (you can query for this string to delete spam or retrieve the list of bugs I've moved)

*** Bug 115982 has been marked as a duplicate of this bug. ***

as far I've seen, this isn't a problem any more more. was just reviewing my votes, and the two sites that were causing this for me now fail with an error ("document contains no data") in the same circumstances they were failing silently before. dupe to bug 19073, on which this depends?

I haven't seen it for ages. WFM? pi

from reading bug 19073, I think that patch (actually attached to a different bug again) was the fix. not sure it's too critical whether it's a dupe or a WFM, as long as it can be closed and confirmed. so... *** This bug has been marked as a duplicate of 19073 ***

REOPEN: On this wire, this is a different error that most of the problem that were duped into bug 19073. Besides, you can't have a depend/block relationship with a dupe. Connection reset could probably be reproduced by restarting the web server during a live transfer. This needs to actually be set up as a testcase, and possibly added to the core networking tests, if not every protocol test, if it is easy to run.

neeti no longer works on mozilla stuff.

marking as fixed. Ben, if you find a test case to exhibit the errors that you postulate, please open a new bug