17732 - stream (result of peer's NewStream method) isn't working

Status: VERIFIED FIXED

(Core Graveyard :: Plug-ins, defect, P3)

Description

[Igor Kushnirskiy]

Overview Description:
When you try to write something to output stream, returned by
PluginInstancePeer::NewStream method, all written data will be lost and
empty result shown/saved.

Steps to reproduce:
1. Apply following patch to mozilla/modules/plugin/test/npsimple.cpp
--
Index: npsimple.cpp
===================================================================
RCS file: /cvsroot/mozilla/modules/plugin/test/npsimple.cpp,v
retrieving revision 1.13
diff -r1.13 npsimple.cpp
44,46c44,46
< #include <gdk/gdk.h>
< #include <gdk/gdkprivate.h>
< #include <gtk/gtk.h>
---
> //#include <gdk/gdk.h>
> //#include <gdk/gdkprivate.h>
> //#include <gtk/gtk.h>
741a742,752
>     // { idk
>     nsIOutputStream *output = NULL;
>
>     if (!NS_FAILED(fPeer->NewStream("text/plain","_new",&output))) {
>         const char * buf = "Hello world";
>         nsresult r;
>         output->Write(buf,strlen(buf),&r);
>         output->Flush();
>         NS_RELEASE(output);
>     }
>     // } idk
829c840
<     if(listener != NULL)
---
>     if(listener != NULL) {
831c842,843
<
---
>         (*listener)->AddRef();
> 	}

--
2. Compile. Run mozilla and open SimpleTest.html

Actual Results:
Sent data will not be shown.

Expected result:

In new window "Hello world" should be shown

Comment 1

[Igor Kushnirskiy]

*** Bug 17543 has been marked as a duplicate of this bug. ***

Comment 2

[Eli Goldberg]

Shrirang is now QA owner for Plug-ins; QA assigning all of my Plug-ins bugs over
to him.

Comment 3

[ekrock's old account (dead)]

Future.

Comment 4

[Peter Lubczynski]

nominating for mozilla1.0, this also effects the NPAPI, but causes a crash! See
the NPAPI tester plugin.

Anyone here have any idea why this is failing?

Comment 5

[av (gone)]

Attachment: patch by peterlubczynski

Comment 6

[av (gone)]

Attachment: patch v.2 -- addresses crash on NPN_DestroyStream and making temp file with no extension so it is not always html

Comment 7

[Peter Lubczynski]

Comment on attachment 62968 [details] [diff] [review]
patch v.2 -- addresses crash on NPN_DestroyStream and making temp file with no extension so it is not always html

very nice! r=peterl

Comment 8

[Peter Lubczynski]

nominating edt0.9.4 because of an important NPAPI compatibility issue and low risk:

1) This fixes NPN_Write
2) Correctly displays the output
3) Fixes CRASH due to bad cast in NPN_DestroyStream which wasn't seen because of #1

Comment 9

[rpotts (gone)]

Comment on attachment 62968 [details] [diff] [review]
patch v.2 -- addresses crash on NPN_DestroyStream and making temp file with no extension so it is not always html

sr=[e-mail]

Comment 10

[av (gone)]

Checked in to the trunk.

Comment 11

[Marek Z. Jeziorek]

marking edt0.9.4+

Comment 12

[av (gone)]

In the branch now.

Comment 13

[av (gone)]

Resolution --> FIXED

Comment 14

[shrirang khanzode]

marking verified.