[Bug 10043] New: Babas Chess version 3.6

http://bugs.winehq.org/show_bug.cgi?id=10043

Summary: Babas Chess version 3.6
           Product: Wine
           Version: 0.9.46.
          Platform: PC
        OS/Version: Linux
            Status: UNCONFIRMED
          Severity: major
          Priority: P2
         Component: test
        AssignedTo: [e-mail]
        ReportedBy: [e-mail]

With wine version 0.9.46 all chat produces an "application error" message box
(does not quit application, only stops messages being sent).

Steps to reproduce: type "tell %USERNAME% message" or type in chat window when
playing.

This did NOT occur with wine versions 0.9.45 and previous.

[Bug 10043] Babas Chess version 3.6

http://bugs.winehq.org/show_bug.cgi?id=10043

Louis Lenders  changed:

What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |[e-mail]
                URL|                            |http://www.babaschess.net/
          Component|test                        |wine-richedit
           Keywords|                            |download, regression

--- Comment #1 from Louis Lenders   2007-10-15 13:08:05 ---
Hi, could you try what happens using native riched20?

I don't have a password/account, and i used version 4.0, so i just tried to
type some text, and i also see the error messagebox. I just typed "qqqqq". Now
with native riched20 i see in the log:

0009:Call KERNEL32.WideCharToMultiByte(00000000,00000000,0033e17c
L"qqqqq",00000006,00000000,00000000,00000000,00000000) ret
=00ea694c
0009:Ret  KERNEL32.WideCharToMultiByte() retval=00000006 ret=00ea694c

Looks ok to me.

With builtin riched20 i see:

0009:Call KERNEL32.WideCharToMultiByte(00000000,00000000,01047228
L"qqqqq\r\n",ffffffff,00f09c00,00000007,00000000,00000000)
 ret=61779a3c
0009:Ret  KERNEL32.WideCharToMultiByte() retval=00000000 ret=61779a3c

and shortly after it "crashes". I guess the problem is the "string" is
mistakenly translated into "string\r\n" by riched20. Maybe some richedit guru
could comment on this?

[Bug 10043] Babas Chess version 3.6

http://bugs.winehq.org/show_bug.cgi?id=10043

Louis Lenders  changed:

What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|UNCONFIRMED                 |NEW
     Ever Confirmed|0                           |1

--- Comment #2 from Louis Lenders   2007-10-15 13:08:34 ---
forgot confirming

[Bug 10043] Babas Chess version 3.6

http://bugs.winehq.org/show_bug.cgi?id=10043

Louis Lenders  changed:

What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |[e-mail]

--- Comment #3 from Louis Lenders   2007-10-15 13:18:05 ---
A quick search through the cvs can be quicker then regression test: i reversed
this patch:

http://www.winehq.org/pipermail/wine-cvs/2007-September/036411.html

And the error messagebox is gone. (I do get a messagebox that i'm not connected
to  a server, but that seems logical)

[Bug 10043] Babas Chess version 3.6

http://bugs.winehq.org/show_bug.cgi?id=10043

--- Comment #4 from Alex Villacís Lasso   2007-10-16 10:33:46 ---
I am the author of the patch at issue.

It is strange, since tests on WinXP-SP2 (provided in a previous patch)
indicated that \r to \r\n transformation by WM_GETTEXT is indeed the correct
behavior for riched20. Maybe this transformation should not be done in a corner
case that BabasChess is triggering. Or maybe this patch, by fixing one
behavior, has uncovered another bug somewhere else - it wouldn't be the first
time.

The problem is that I am having trouble running the application at all. My
attempts fail because of a call to an unimplemented stub in gdiplus.dll, before
reaching any edit box. This means that the submitter of the bug is probably
running with a native gdiplus.dll. I am still looking for a suitable native
gdiplus.dll, since my installation of WinXP-SP2 does not provide it.

Could the author of the bug please run the app with builtin riched20, *without*
reversing the riched20 patch, and with WINEDEBUG=+seh,+riched20, and post the
resulting trace?

What happens with a clean installation of Wine? Does the application run at
all? (It does *not* in my case...)

[Bug 10043] Babas Chess version 3.6

http://bugs.winehq.org/show_bug.cgi?id=10043

--- Comment #5 from Louis Lenders   2007-10-16 10:48:52 ---
Hi Alex, i'm not the reporter of the bug, but for me it runs just fine without
native gdiplus. I used version 4.0 however:
http://www.babaschess.net/download/SetupBabasChess_4_0_XP.exe

but i guess it behaves more or less the same as version 3.6. Could you try the
application from the link above, and see if you can get it started. Regards

[Bug 10043] Babas Chess version 3.6

http://bugs.winehq.org/show_bug.cgi?id=10043

--- Comment #6 from Alex Villacís Lasso   2007-10-16 11:01:01 ---
(In reply to comment #5)
> Hi Alex, i'm not the reporter of the bug, but for me it runs just fine without
> native gdiplus. I used version 4.0 however:
> http://www.babaschess.net/download/SetupBabasChess_4_0_XP.exe
>
> but i guess it behaves more or less the same as version 3.6. Could you try the
> application from the link above, and see if you can get it started. Regards

Nope, still no luck. That was the version I tried myself:

[alex@srv64 BabasChess]$ ls
BabasChess.chm      BabasChess.exe      BabasChess_NLD.MLF    Data
BabasChess_CRO.MLF  BabasChess_FIN.MLF  BabasChess_POL.MLF    DefLayouts
BabasChess_DEU.MLF  BabasChess_FRA.MLF  BabasChess_POR.MLF    Engines
BabasChess_ENG.MLF  BabasChess_HUN.MLF  BabasCrashReport.exe  Themes
BabasChess_ESP.MLF  BabasChess_ITA.MLF  CustomCmd             timeseal.exe
[alex@srv64 BabasChess]$ ./BabasChess.exe 
ALSA lib conf.c:3949:(snd_config_expand) Unknown parameters 0
ALSA lib control.c:909:(snd_ctl_open_noupdate) Invalid CTL default:0
fixme:richedit:RichEditWndProc_common WM_STYLECHANGING: stub
fixme:richedit:RichEditWndProc_common WM_STYLECHANGED: stub
fixme:richedit:RichEditWndProc_common WM_STYLECHANGING: stub
fixme:richedit:RichEditWndProc_common WM_STYLECHANGED: stub
fixme:richedit:RichEditWndProc_common WM_STYLECHANGING: stub
fixme:richedit:RichEditWndProc_common WM_STYLECHANGED: stub
fixme:richedit:RichEditWndProc_common ECO_AUTOWORDSELECTION not implemented
yet!
fixme:richedit:RichEditWndProc_common ECO_AUTOVSCROLL not implemented yet!
fixme:richedit:RichEditWndProc_common ECO_NOHIDESEL not implemented yet!
fixme:richedit:RichEditWndProc_common EM_SETTARGETDEVICE: stub
wine: Call from 0x7b840ab0 to unimplemented function
gdiplus.dll.GdipDrawImageRectRectI, aborting

All done with current git, brand new .wine (deleted previous .wine &&
wineprefixcreate). No other configuration performed after wineprefixcreate
except run setup of BabasChess, then attempt to run the app itself. Crash
happens right after answering dialog about creating personalized directory
(left at default path).

[Bug 10043] Babas Chess version 3.6

http://bugs.winehq.org/show_bug.cgi?id=10043

--- Comment #7 from Louis Lenders   2007-10-16 11:19:08 ---
yeah, you are right, guess i had a native gdiplus lying in system32 somewhere

You could try 'wget http://kegel.com/wine/winetricks && sh winetricks gdiplus'

that should get you around that problem

[Bug 10043] Babas Chess version 3.6

http://bugs.winehq.org/show_bug.cgi?id=10043

--- Comment #8 from Alex Villacís Lasso   2007-10-18 10:55:49 ---
Created an attachment (id=8643)
 --> (http://bugs.winehq.org/attachment.cgi?id=8643)
Patch to fix behavior of VK_RETURN in single-line richedit control

As I thought, this is a case of a bugfix uncovering another bug.

The problematic controls are single-line richedit controls at the bottom of the
application window. The user is supposed to type text and hit RETURN (WM_CHAR =
VK_RETURN) in order to send a message. The application sends a WM_GETTEXT
message in order to retrieve the text the user just entered, and process it.

The behavior of the single-line richedit control in builtin was never right to
begin with. For the sake of example, let's say the user types "qqq", then
presses ENTER. On a single-line richedit control, native riched20 returns "qqq"
without any \r or \n tacked on. This indicates that the control refuses to
insert carriage returns into the text of a single-line richedit control. On the
same scenario, the old code in builtin riched20 returned "qqq\r", which is
already wrong (the hidden bug). However, this was not enough (yet) to break
BabasChess.

The new code in builtin returns "qqq\r\n", because (as demonstrated by tests),
any carriage returns must be converted to \r\n pairs when retrieved via
WM_GETTEXT. However, since this is a single-line richedit control, there was
not supposed to be any carriage returns in the first place. With the extra \n
added, BabasChess broke, as experienced by the submitter of the bug.

The attached patch fixes this by enforcing the rule that WM_CHAR may not insert
carriage returns (VK_RETURN) into the text of a single-line richedit control,
and by adding a test to check for this behavior. This patch will shortly be
sent to wine-patches.

[Bug 10043] Babas Chess version 3.6

http://bugs.winehq.org/show_bug.cgi?id=10043

--- Comment #9 from Alex Villacís Lasso   2007-12-03 11:58:31 ---
Patch has been committed to current git. Please confirm whether bug has been
resolved.

[Bug 10043] Babas Chess version 3.6

http://bugs.winehq.org/show_bug.cgi?id=10043

[e-mail] changed:

What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|                            |FIXED

--- Comment #10 from [e-mail]  2007-12-03 18:54:58 ---
Sorry it took so long to mark as fixed. The patch works here.

[Bug 10043] Babas Chess version 3.6

http://bugs.winehq.org/show_bug.cgi?id=10043

--- Comment #11 from Louis Lenders   2007-12-04 10:23:27 ---
so it;s fixed

[Bug 10043] Babas Chess version 3.6

http://bugs.winehq.org/show_bug.cgi?id=10043

Louis Lenders  changed:

What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|RESOLVED                    |CLOSED

--- Comment #12 from Louis Lenders   2007-12-04 10:23:47 ---
closing