wine-devel

wine-devel@winehq.org

3 participants
84400 discussions

Re: none
by Alexandre Julliard 26 Oct '01

26 Oct '01

Andriy Palamarchuk <apa3a(a)yahoo.com> writes: > Ok, how about this: > Each system parameter has "is_loaded" flag which shows > whether the parameter is loaded or not. Parameters, > stored directly in SysParametersInfo are loaded from > registry only when they are requested and the flag is > "false", after the loading the flag is set to "true". > On interprocess notification of the parameter change > the corresponding parameter flag is set to "false". > The parameter will be reloaded when it is requested > the next time. Yes, something like this would work. > Are there any existing implementations of > communication between Wine processes I can use as an > example? You want to use some variant of SendMessage in this case, probably a broadcast SendNotifyMessage with an internal message code (adding a WM_WINE_ case in dlls/user/message.c). > Do you still think it is better don't affect X > settings from Wine? Yes. > While we are on this - what about other my patch - > "hello5.c"? It fixes a warning and an error message I > got when I tried to compile hello5.c. The error depends on the compiler version. You should check the DUMMYUNIONNAME defines if you want to make it work for all platforms. I'm not sure it's worth the trouble though. -- Alexandre Julliard julliard(a)winehq.com

1 0

(no subject)
by Andriy Palamarchuk 26 Oct '01

26 Oct '01

Alexandre Julliard wrote: > Andriy Palamarchuk <apa3a(a)yahoo.com> writes: > > >>Do you think it is better to do lazy loading for most >>parameters? >> > > Yes definitely. Ok, how about this: Each system parameter has "is_loaded" flag which shows whether the parameter is loaded or not. Parameters, stored directly in SysParametersInfo are loaded from registry only when they are requested and the flag is "false", after the loading the flag is set to "true". On interprocess notification of the parameter change the corresponding parameter flag is set to "false". The parameter will be reloaded when it is requested the next time. Parameters, backed by GetSystemMetrics are processed differently. These parameters are always loaded. They are reloaded right away on interprocess change notification. We can't do lazy loading because GetSystemMetrics returns its values from system metrics array and we don't want to complicate this function. I'm going to implement the flags as a bit array. Index constants for this array will be defined separately. >>I'll think about it more. Originally I planned to >>broadcast change notification across processes and use >>the reload function in other process to refresh values >>from registry. The problem with this approach is when >>a parameter is changed without saving. >> > > A possibility could to create a volatile registry branch to save > parameters that aren't permanently saved. This way they would be > accessible from other processes. Thank you for the idea. Are there any existing implementations of communication between Wine processes I can use as an example? Disregard the patches "SystemParametersInfo / GetSystemMetrics", "SPI_GET/SETBORDER". I'll implement the discussed above changes and resubmit them. Do you still think it is better don't affect X settings from Wine? Please let me know. While we are on this - what about other my patch - "hello5.c"? It fixes a warning and an error message I got when I tried to compile hello5.c. Andriy Palamarchuk __________________________________________________ Do You Yahoo!? Make a great connection at Yahoo! Personals. http://personals.yahoo.com

1 0

ReadFile behaviour on COMs
by Uwe Bonnes 26 Oct '01

26 Oct '01

Hallo, somebody showed me an application speaking to a device connected to the serial port. The application failed after some successfull communication steps. As the protocoll is available in some other (open source programm) we could see that the program requested two bytes from the device , but ReadFile only delivered one byte (with that byte beeing the right one of the expected two byte sequence). I suspect ReadFile to be the culprit. While the api description tells to read all requested bytes ( beside of real errors), our implementation happilies returns fewer bytes than requested. In the case of COM devices, all timeouts of the SetCommTimeout structure should be checked two. Anyone with an implementaion for that? Bye -- Uwe Bonnes bon(a)elektron.ikp.physik.tu-darmstadt.de Institut fuer Kernphysik Schlossgartenstrasse 9 64289 Darmstadt --------- Tel. 06151 162516 -------- Fax. 06151 164321 ----------

1 0

(no subject)
by Mike McCormack 26 Oct '01

26 Oct '01

Hi Andreas, i had a bit more free time, so i tracked down the rendering problem in NetTerm a bit more. The problem occurs after SelectVisRgn16 is called with the same region as that currently selected in the DC. (eg. dc->hVisRgn == hrgn ). I made a dodgy patch to fix the problem by incrementing the reference count of the region in this case, but i don't know how to fix it properly, because i don't know much about region handling. With the dodgy patch, NetTerm's windows paints properly. Perhaps perhaps you or somebody has a better way to fix the problem? Mike Andreas Mohr wrote: > On Tue, Oct 09, 2001 at 11:35:39AM +1000, Mike McCormack wrote: > > > > Hi All, > > > > i'm trying to solve a comms problem in NetTerm > > http://www.securenetterm.com/ but i've run into another problem. > > > > The rendering in NetTerm fails almost completely. I > > Yep, I strongly suspect a reference count problem here, too. > ------------------------------------------ mailto:Mike_McCormack@start.com.au ph +82 16 430 0425 __________________________________________________________________ Get your free Australian email account at http://www.Looksmart.com.au

1 0

Re: cursor visibility problem
by Alexandre Julliard 25 Oct '01

25 Oct '01

Ove Kaaven <ovehk(a)ping.uio.no> writes: > In Alexandre's new windowing architecture, we have in > dlls/x11drv/window.c, line 119, this: > > attr->cursor = X11DRV_GetCursor( display, GlobalLock16(GetCursor()) ); > > Problem: what if ShowCursor(FALSE) is called to hide the cursor *before* > the window was created? Yup, GetCursor still returns the current (arrow) > cursor, so the new window gets associated with a visible cursor, though > the app would expect the cursor to remain hidden. Yes, cursor handling has some problems right now. It also needs to be a lot more clever about setting the cursor on child windows, and not walk the full list of top-level windows on every SetCursor. It will be fixed eventually, but it's not my top priority right now. -- Alexandre Julliard julliard(a)winehq.com

1 0

cursor visibility problem
by Ove Kaaven 25 Oct '01

25 Oct '01

In Alexandre's new windowing architecture, we have in dlls/x11drv/window.c, line 119, this: attr->cursor = X11DRV_GetCursor( display, GlobalLock16(GetCursor()) ); Problem: what if ShowCursor(FALSE) is called to hide the cursor *before* the window was created? Yup, GetCursor still returns the current (arrow) cursor, so the new window gets associated with a visible cursor, though the app would expect the cursor to remain hidden. Diablo II is currently known to suffer from this problem... it has too many cursors visible. I haven't come up with a very clean solution to this... ideas?

1 0

winelib works only in debugmsg mode
by Guilherme Kunz 25 Oct '01

25 Oct '01

Hi, I have compiled a program with mfc using winemaker and when a run the program "./dlg" show a messagebox "Could not get address of WinMain", i am linking the mfc static mode. but when i run "./dlg --debugmsg +all &> debug.log" the program run right. gcc 2.96.85 kernel 2.2.19 Any idea ?? Thank you. Guilherme Kunz

1 0

Re: wine/memory virtual.c
by Alexandre Julliard 25 Oct '01

25 Oct '01

Ove Kaaven <ovehk(a)ping.uio.no> writes: > Well, though I haven't tested it yet (working on other stuff right now), > if it works for you, it'll work for me. I haven't really tested it either, but I'll commit it and we'll see... -- Alexandre Julliard julliard(a)winehq.com

1 0

Re: wine/memory virtual.c
by Alexandre Julliard 25 Oct '01

25 Oct '01

Alexandre Julliard <julliard(a)winehq.com> writes: > It's used at least in debug events, so yes it's possible that the > debugger will try to use the handle to read the file. Probably better > to close it completely when on removable media; and we'll need to do > this for dlls too. I'll try to come up with something. OK how about this? Index: loader/pe_image.c =================================================================== RCS file: /opt/cvs-commit/wine/loader/pe_image.c,v retrieving revision 1.99 diff -u -r1.99 pe_image.c --- loader/pe_image.c 2001/10/02 17:49:20 1.99 +++ loader/pe_image.c 2001/10/24 19:39:21 @@ -667,6 +667,12 @@ if (nt->FileHeader.Characteristics & IMAGE_FILE_DLL) { + if (hFile) + { + UINT drive_type = GetDriveTypeA( wm->short_filename ); + /* don't keep the file handle open on removable media */ + if (drive_type == DRIVE_REMOVABLE || drive_type == DRIVE_CDROM) hFile = 0; + } SERVER_START_REQ( load_dll ) { req->handle = hFile; Index: scheduler/process.c =================================================================== RCS file: /opt/cvs-commit/wine/scheduler/process.c,v retrieving revision 1.162 diff -u -r1.162 process.c --- scheduler/process.c 2001/10/03 18:40:10 1.162 +++ scheduler/process.c 2001/10/24 19:39:22 @@ -327,6 +327,17 @@ lstrcpynA( main_exe_name, full_argv0, sizeof(main_exe_name) ); } + if (main_exe_file) + { + UINT drive_type = GetDriveTypeA( main_exe_name ); + if (drive_type == DRIVE_REMOVABLE || drive_type == DRIVE_CDROM) + { + /* don't keep the file handle open on removable media */ + CloseHandle( main_exe_file ); + main_exe_file = 0; + } + } + /* Retrieve entry point address */ entry = (LPTHREAD_START_ROUTINE)((char*)current_process.module + PE_HEADER(current_process.module)->OptionalHeader.AddressOfEntryPoint); Index: server/process.c =================================================================== RCS file: /opt/cvs-commit/wine/server/process.c,v retrieving revision 1.70 diff -u -r1.70 process.c --- server/process.c 2001/07/14 00:50:30 1.70 +++ server/process.c 2001/10/24 19:39:22 @@ -788,8 +788,9 @@ /* signal the end of the process initialization */ DECL_HANDLER(init_process_done) { - struct file *file; + struct file *file = NULL; struct process *process = current->process; + if (!process->init_event) { fatal_protocol_error( current, "init_process_done: no event\n" ); @@ -798,11 +799,10 @@ process->exe.base = req->module; process->exe.name = req->name; - if (req->exe_file && (file = get_file_obj( current->process, req->exe_file, GENERIC_READ ))) - { - if (process->exe.file) release_object( process->exe.file ); - process->exe.file = file; - } + if (req->exe_file) file = get_file_obj( current->process, req->exe_file, GENERIC_READ ); + if (process->exe.file) release_object( process->exe.file ); + process->exe.file = file; + generate_startup_debug_events( current->process, req->entry ); set_event( process->init_event ); release_object( process->init_event ); -- Alexandre Julliard julliard(a)winehq.com

2 1

ddraw fullscreen-app problem in latest CVS
by TAKESHIMA Hidenori 25 Oct '01

25 Oct '01

I can run some full-screen app in 2001/10/20 cvs. but, in the latest cvs, wine may crash in creating the fullscreen-primary surface... (I don't know why, but since wine-server protocol has been updated?????) ddrawtst.c is a minimal test program that crashes the latest cvs. My environment is: debian2.3-woody, XFree86-4.1.0-7. No permission to access /dev/kmem directly.

1 0

Jump to page:

2025

2024

2023

2022

2021

2020

2019

2018

2017

2016

2015

2014

2013

2012

2011

2010

2009

2008

2007

2006

2005

2004

2003

2002

2001

wine-devel