July 2002 - Wine-devel - list.winehq.org

The filedlg95 behavior
by Duane Clark 06 Jul '02

06 Jul '02

I was hoping to perhaps provoke discussion of the new behavior of the file dialog. Prior to this patch: http://www.winehq.com/hypermail/wine-cvs/2002/06/0028.html the default behavior was to start the open file dialog in the current directory. After the patch, the default behavior is to start in the C:\My Documents directory, though this can be changed with a registry key. While this is the default behavior on Windows, I am not sure that wine should copy this behavior so precisely. I personally think that the old behavior was better. I would think that very few Linux users would actually use the C:\My Documents directory. But perhaps my own preferences are coloring my views. I think the command line oriented nature of Linux would call for a occasional differences in behavior. Duane

2 2

RegEnumValueA bugs and patch in dlls/advapi32/registry.c
by Max 05 Jul '02

05 Jul '02

Well, finally found and corrected a bug that made AutoCAD R14 hang... I guess some other apps could have the same problem solved, but had no time to test more :) In RegEnumValueA, when the size for requested data buffer is too small, function should return an error (as it does), but also return correct buffer size (as it doesn't). Found in AutoCAD; it asked for a reistry key value, got error, and then looped forever trying to get val with wrong buffer size. I guess, but had no time yet to check, that this problem could happen somwhere else in Registry.c, I'll check it out next days. With this patch, finally AutoCAD R14 is perfectly usable in wine, even without a native windows installed ! Here the patch : --- dlls/advapi32/registry.c Sat Jul 6 01:29:08 2002 +++ /home/massimo/wine-patches/registry.c Sat Jul 6 01:35:32 2002 @@ -1157,6 +1157,9 @@ RtlUnicodeToMultiByteSize( &len, info->Name, info->NameLength ); if (len >= *val_count) { + /* CHANGELOG : 04.07.2002 by Max + MUST RETURN REQUESTED CORRECT BUFFER SIZE, NOT ONLY TELL IT'S WRONG */ + *val_count = len + 1 ; status = STATUS_BUFFER_OVERFLOW; goto done; } @@ -1174,6 +1177,9 @@ { if (len > *count) { + /* CHANGELOG : 04.07.2002 by Max + MUST RETURN REQUESTED CORRECT BUFFER SIZE, NOT ONLY TELL IT'S WRONG */ + *count = len+1 ; status = STATUS_BUFFER_OVERFLOW; goto done; } I'll just post it to wine-patches too. Regards Max

1 0

Re: Floppy handling Fix
by Alexandre Julliard 05 Jul '02

05 Jul '02

Keith Matthews <keith_m(a)sweeney.demon.co.uk> writes: > 1 - Fix incorrect reporting of no media in floppy drive as > permissions problem. > 2 - Provide partial workaround for very long delays (c. 40 sec per > lookup, about 7 minutes to set up file browser) on lookups to > floppy drive not physically present (e.g. laptop with floppy or > CD setup) I'm not sure why you are trying to duplicate the kernel permission checks, this is clearly wrong. You should simply do an open and check errno to find out what went wrong. Also there's no point in having a DRIVE_NO_MEDIA flag since this is a temporary condition. For the case of open succeeding with no floppy drive, I think you could handle it by checking if the subsequent read() fails. -- Alexandre Julliard julliard(a)winehq.com

1 0

Re: Better fix for [bug 829] (same surface display corruption)
by Lionel Ulmer 05 Jul '02

05 Jul '02

> I took Lionel's sugestions to heart and came up with a better fix for > bug 829. This has a better check for defining when the corruption could > occur and should be a little faster Well, I had not in mind exactly what you did. > + SameSurfaceOK = (!((src == iface)&&(sbase < dbuf)&&(xdst.top < xsrc.bottom)&&(xdst.left < xsrc.right))); I do not really understand all the details of this check, so a comment would be nice, but well, I think it is unneeded (see below). > + if (SameSurfaceOK) { > + for (y = 0; y < dstheight; y++) { > + memcpy(dbuf, sbuf, width); > + sbuf += sdesc.u1.lPitch; > + dbuf += ddesc.u1.lPitch; > + } > + } else { > + > + sbuf += (sdesc.u1.lPitch*dstheight); > + dbuf += (ddesc.u1.lPitch*dstheight); > + > + for (y = 0; y < dstheight; y++) { > + sbuf -= sdesc.u1.lPitch; > + dbuf -= ddesc.u1.lPitch; > + memcpy(dbuf, sbuf, width); > + } > + } > + The problem I can see is that you use 'memcpy' even in the SameSurfaceNotOK case. And this is bad as memcpy does NOT handle overlapping memory zones. Moreover, you only have the 'descending' copy order in that case and not both cases. What I had in mind in my comments to your patch was something like that : if (src != iface) Do the standard memcpy code path else if source_Y > dest_Y do the ascending memmove copy else do the descending memmove copy This way you did not hurt the common case (ie Blt from two different surfaces). Now I agree that even in the same surface case, one could add some checks to see if the rectangles overlap or not, but I do not know if it would be worth the pain as memmove should not be THAT slower than memcpy. Lionel -- Lionel Ulmer - http://www.bbrox.org/

3 3

RE: warcraft 3
by Hans Leidekker 05 Jul '02

05 Jul '02

> After buying Warcraft 3, and reading on the mailing list that the beta had > worked under wine, I decided to try the final version under wine. It installed > fine, but won't run. Or rather, when I try to run it, it prints this out, and > just sits there: > > fixme:cdrom:CDROM_GetIdeInterface CD-ROM device with major ID 11 not supported > fixme:cdrom:CDROM_GetIdeInterface CD-ROM device with major ID 11 not supported > fixme:midi:OSS_MidiInit Synthesizer support MIDI in. Not supported yet (please report) Let me guess: you have SCSI cdrom drive. I stumbled on this one as well some time ago. I learned that 11 is the SCSI major. It looks strange to me that the function CDROM_GetIdeInterface gets called while we have a SCSI cdrom drive? Anyways, as a workaround -- I don't think this is the right solution -- I added the following line to the switch statement in cdrom:CDROM_GetIdeInterface: case SCSI_CDROM_MAJOR: *iface = 0; break; This allowed my application to proceed (don't remember which application that was though). Regards, -Hans Leidekker

2 1

GRand Prix Legends network code
by Christoph Frick 05 Jul '02

05 Jul '02

Hi! I did some debugging, to find out why the game Grand Prix Legends did not work as network client/server. The problem: Its only possible to connect to servers/let clients connect from the local network. Solution: The game set (WS_)SO_DONTROUTE on the socket - and of course this works perfect under unix - but i never had problems while i played this game under windows so this is either not set, when the game runs under windows - or it has no effect within windows. Also the game like to bind the socket for the client connection a second time, when the first connection failed or was canceled. This does not work (addr. in use) - so a (WS_)SO_REUSEADDR would have been needed. Could please someone with a deeper insight in the world of windows direct me to some docu/specs or what ever to get informations about the winsock stuff? The link in the header of the dlls/winsock/socket.c is not valid. I even dont know what winsock version is the current one (1.1,2.0,???). Help is appreciated! Thanx alot! CU!

1 0

Re: dlls/advapi32/registry.c patch for RegEnumValuesA
by Massimo 05 Jul '02

05 Jul '02

Resubmitting as for Tony Lambregts request :) Well, finally found and corrected a bug that made AutoCAD R14 hang... I guess some other apps could have the same problem solved, but had no time to test more :) In RegEnumValueA, when the size for requested data buffer is too small, function should return an error (as it does), but also return correct buffer size (as it doesn't). Found in AutoCAD; it asked for a reistry key value, got error, and then looped forever trying to get val with wrong buffer size. I guess, but had no time yet to check, that this problem could happen somwhere else in Registry.c, I'll check it out next days. With this patch, finally AutoCAD R14 is perfectly usable in wine, even without a native windows installed ! Change log: Change In RegEnumValueA, when the size for requested data buffer is too small, function should return an error (as it did), and also return correct buffer size (now it does). Max max(a)com.veneto [reversed :) ] ----- --- dlls/advapi32/registry.c Sat Jul 6 01:29:08 2002 +++ /home/massimo/wine-patches/registry.c Sat Jul 6 01:35:32 2002 @@ -1157,6 +1157,9 @@ RtlUnicodeToMultiByteSize( &len, info->Name, info->NameLength ); if (len >= *val_count) { + /* MUST RETURN REQUESTED CORRECT BUFFER SIZE, NOT ONLY TELL IT'S WRONG */ + *val_count = len + 1 ; status = STATUS_BUFFER_OVERFLOW; goto done; } @@ -1174,6 +1177,9 @@ { if (len > *count) { + /* MUST RETURN REQUESTED CORRECT BUFFER SIZE, NOT ONLY TELL IT'S WRONG */ + *count = len+1 ; status = STATUS_BUFFER_OVERFLOW; goto done; }

2 1

Re: Added support for APPMODE=<type>[:<mainFunc>]
by Alexandre Julliard 05 Jul '02

05 Jul '02

Gregg Mattinson <gm138242(a)scot.canada.sun.com> writes: > - .spec files can't be compiled with -g when using Forte C. This causes static > variables to have a "random" prefix attached to their name, and prevents the > assembly code from linking to them correctly. > - Forte C has linker issues, so cui executables with main() conflicted with > main() in $WINELOADER. Renaming main() to main2() solved the issue, so I added > support to winebuild to allow the main function name to be optionally specified > in the makefile. I have applied the rest of the patch, but these are bugs that need to be fixed, not simply worked around. -- Alexandre Julliard julliard(a)winehq.com

1 0

Re: dlls/advapi32/registry.c patch for RegEnumValuesA
by Tony Lambregts 04 Jul '02

04 Jul '02

Max wrote: >Here a patch that corrects a nasty bug in dll/advapi32/registry.c, causing >AutoCAD R14 (and I guess other apps too...) to hang. > >Regards > >Max > > > >------------------------------------------------------------------------ > >--- dlls/advapi32/registry.c Sat Jul 6 01:29:08 2002 >+++ /home/massimo/wine-patches/registry.c Sat Jul 6 01:35:32 2002 >@@ -1157,6 +1157,9 @@ > RtlUnicodeToMultiByteSize( &len, info->Name, info->NameLength ); > if (len >= *val_count) > { >+ /* CHANGELOG : 04.07.2002 by Max >+ MUST RETURN REQUESTED CORRECT BUFFER SIZE, NOT ONLY TELL IT'S WRONG */ > The comment is OK I suppose, but it is part of the change log. Change log entries should not be part of the patch itself. >+ *val_count = len + 1 ; > status = STATUS_BUFFER_OVERFLOW; > goto done; > } >@@ -1174,6 +1177,9 @@ > { > if (len > *count) > { >+ /* CHANGELOG : 04.07.2002 by Max >+ MUST RETURN REQUESTED CORRECT BUFFER SIZE, NOT ONLY TELL IT'S WRONG */ > Same here. >+ *count = len+1 ; > status = STATUS_BUFFER_OVERFLOW; > goto done; > } > > Sorry to nitpick could you resubmit something like this ---- This is the first part. This is where you can say what you did, and say what you wanted to say, in as long and as rambling way as you want. It is for expaining whatever you wanted to expain, why you wanted to do it the way you wanted to do it section. next comes the change log . after that your name and contact. Change log: Change In RegEnumValueA, when the size for requested data buffer is too small, function should return an error (as it did), and also return correct buffer size (now it does). Max max(a)veneto.co ----- attched patch

1 0

Running half-life turns X display into garbage
by Joshua Thielen 04 Jul '02

04 Jul '02

Hello, When I run half-life, the X display gets totally corrupted and the only way to restore is to restart the X server. I'm using an NVidia G-force 3 video card running at 1024x768 with 16-bit color. I have already run a few traces, but since I have no idea at what point the display gets corrupted, it's kind of hard to debug. Has anyone encountered this problem yet? Are there any specific functions or channels I should be looking at in the trace? Thanks, Josh Thielen __________________________________________________ Do You Yahoo!? Sign up for SBC Yahoo! Dial - First Month Free http://sbc.yahoo.com

4 4