101514 - [FIX]crash when cancelling out of Print dialog

Status: VERIFIED DUPLICATE of bug 101221

(Core :: Printing: Output, defect, P1)

Description

[rods (gone)]

Comment 1

[rods (gone)]

I can't figure out when this started or why it started.

Comment 2

[rods (gone)]

It start with mkaply's check in 1.148 of the nsDocumentViewer.cpp for 
[e-mail].

The problem is these lines:
    factory->CreateDeviceContextSpec(mWindow, devspec, aSilent);
    if (nsnull != devspec) {

got changed to:
    rv = factory->CreateDeviceContextSpec(mWindow, devspec, aSilent);
    if (NS_SUCCEEDED(rv)) {

And on Window the CreateDeviceContextSpec call is not passing back error codes.

THIS SHOULD HAVE BEEN TESTED ON Windows before being checked in, Don should have 
done that before giving it an r=

The solution is for the factory to pass back the correct error codes which will 
also help with Bug 101289.

Here is the patch:

Index: nsDeviceContextSpecFactoryW.cpp
===================================================================
RCS file: /cvsroot/mozilla/gfx/src/windows/nsDeviceContextSpecFactoryW.cpp,v
retrieving revision 3.20
diff -u -r3.20 nsDeviceContextSpecFactoryW.cpp
--- nsDeviceContextSpecFactoryW.cpp     2001/09/18 22:02:30     3.20
+++ nsDeviceContextSpecFactoryW.cpp     2001/09/25 12:24:12
@@ -653,7 +653,11 @@
       //don't free the DEVMODE because the device context spec now owns it...
       ::GlobalUnlock(prntdlg.hDevNames);
       ::GlobalFree(prntdlg.hDevNames);
+    } else {
+      rv = NS_ERROR_FAILURE;
     }
+  } else {
+    rv = NS_ERROR_ABORT;
   }

   return rv;

Comment 3

[Roland Mainz]

> THIS SHOULD HAVE BEEN TESTED ON Windows before being checked in, Don should 
> have done that before giving it an r=

The patch was tested on windows (printing itself works), however noone of us
tested the "cancel the print dialog" codepath on windows (I don't have a Win32
build machine myself)... looks I blindly trusted that the tester checked that,
too ...
_SORRY_ ... ;-(

Comment 4

[Roland Mainz]

Uhm... isn't this a DUPLICATE of bug 101221 ("regression: cancelling printing a
page crashes - Trunk [@ nsDeviceContextWin::GetDeviceContext") ?

Comment 5

[rods (gone)]

Roland, thanks for the dup, I am behind on looking at the bugs I am copied on.

*** This bug has been marked as a duplicate of 101221 ***

Comment 6

[Roland Mainz]

rods:
bug 101221 has already a sr=kin, I only need an r= from you and someone who
checks the patch "in", please ...

Comment 7

[timeless]

vrfy dupe