144406 - Attempt to read attachment file from bad media results in message sent with corrupt attachment

Status: RESOLVED DUPLICATE of bug 74632

(MailNews Core :: Composition, defect)

Description

[Pádraig Brady]

From Bugzilla Helper:
User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:0.9.8) Gecko/20020205
BuildID:    2002020511

Note this happened on RC2, not the above build ID (0.9.9).
Anyway I attached a file from a floppy disk to a mail,
but mozilla didn't complain even though it could only
attach the first 16K of the file!

Reading the file using cat gave an I/O error, so
mozilla should abort a file if it gets an error from read()

Reproducible: Always
Steps to Reproduce:
1.Hard bit is to get file on floppy that gives I/O error
2.attach to mail


Actual Results:  attaches any part of file read, even if read() returns error

Expected Results:  Display error, and don't attach corrupt file!

Comment 1

[Pádraig Brady]

Hmm just noticed 74632 which is very much related,
though I think the problem there is read() blocks 
forever trying to read files when sending?

Comment 2

[Eric Coleman]

corrupt email attachments on win2k as well.  Everything that gets sent is corrupt.

Comment 3

[Trix Supremo]

changing qa contact to yulian

Comment 4

[Mike Cowperthwaite]

I've reproduced this with TB 1.5-0215, Windows 2000.  It's not really an "attachments" issue; that category is for handling attachments on mail that's 
in the Inbox, not for sending.

In my case, the file on the floppy disc is reported (when copying in Windows Explorer) as having a bad CRC.  When I select the file for attachment, there is a long period of time spent accessing the floppy, then the file-select dialog closes and the file appears in the attachment bucket (name correct, generic Application icon).  When I send the file, there is additional floppy access 
(but not as much -- the corrupted attachment, when saved to disk, is only 32KB, compared to the 1.2 MB of the original file), then the message is sent.

Comment 5

[Wayne Mery (:wsmwk)]

Mike, can you still repro?  If still exist, any reason to not simply dup to bug 74632?  although bug 74632 is somewhat lacking info, espicially after comment 2