September 2001 - Wine-Devel

Re: ukrainian layout
by Dmitry Timoshkov Sept. 8, 2001

Sept. 8, 2001

"Andy Rysin" <arysin(a)yahoo.com> wrote: > Also I had to modify unicode/cp20866.c (KOI8-R) to allow conversion > of additional 4 pair of letters in Ukrainian language (KOI8-U, > RFC2319) as well as one additional pair of letters for Belorussian > language (AKA KOI8-RU, which does not have RFC though). > Strictly speaking KOI8-R is a subset (if not to take to account > pseudographics) of KOI8-U wich is a subset of KOI8-RU. > That's why I applied it to cp20866.c instead of creating smth new. > (I took it from my patches for linux kernel nls system) You can't modify cp*.c files, they are auto generated from the unicode.org file set. You need to send your patch to the unicode.org maintainers and hopefully it will be included in the next Unicode Standard release. > Another problem is that I could not test it practiacally because for > some reason Win32 programs under wine does not input cyrillic > letters, but if I swicth to X program and then back it stops input at > all if layout is cyrillic. If I swicth back to latin everything's ok. It's a known problem, that Wine's x11 keyboard driver does not work with "non-sticky" switches. I have tested cyrillic input with XFree86 3.3.6, 4.0.3 and 4.1.0 with KDE and KiKbd, all works fine for me. Proper support for non-sticky modifiers, dead-keys and the real keyboard layout support requires almost complete keyboard driver rewrite. > BTW, why not to take layouts from X system itself? > Those wich are under /etc/X11/xkb/symbols Because XFree86 is not the only X11 implementation out there. -- Dmitry.

2 2

Re: Make cmdline writable (cmdline4)
by Alexandre Julliard Sept. 8, 2001

Sept. 8, 2001

Francois Gouget <fgouget(a)free.fr> writes: > In the case below. Works on Win98 but not in Wine. I believe the > reason is that Windows does not need to modify the command line anyway. In some cases it does; for instance your test program should fail under NT when using Unicode. Do you have a real app that requires this? -- Alexandre Julliard julliard(a)winehq.com

3 2

WM_[LR]BUTTON{DOWN,DBLCLK} in menus
by Bobby Bingham Sept. 8, 2001

Sept. 8, 2001

I was writing a stupid app to help my friend learn to use menus in windows apps. It would call TrackPopupMenu on a WM_LBUTTONDOWN message. After some experimentation, I found that if I leftclicked in the client are, the menu would appear, as it should. But if, while the menu was still active, I leftclicked in the window's nonclient area (titlebar), the menu would appear elsewhere. By clicking the menu receives a WM_LBUTTONDOWN message, and determines that it should close the menu as it's not inside the menu, and it doesn't remove the message. This is normally fine, but if the click falls on the nonclient area of a window, the message should not be passed through. I believe this also applies to WM_LBUTTONDBLCLK, and the right button equivalents. Is there an easy way to determine what window the click would be passed through to, so we can check if it'll be in the client area or not? Bobby Bingham -- Microsoft Windows: Made for the internet The Internet: Made for UNIX

1 0

Re: Problems with WM_PAINT and threads
by Alexandre Julliard Sept. 7, 2001

Sept. 7, 2001

Ove Kaaven <ovehk(a)ping.uio.no> writes: > After Wine dispatches this WM_PAINT message from the wrong thread, > 3DMark2000 fails to locate its thread-local data, gets confused, and > crashes. Does this help? Index: windows/win.c =================================================================== RCS file: /opt/cvs-commit/wine/windows/win.c,v retrieving revision 1.147 diff -u -r1.147 win.c --- windows/win.c 2001/08/29 00:16:00 1.147 +++ windows/win.c 2001/09/07 19:34:24 @@ -387,15 +387,17 @@ hwndRet = pWnd->hwndSelf; /* look among siblings if we got a transparent window */ - while (pWnd && ((pWnd->dwExStyle & WS_EX_TRANSPARENT) || - !(pWnd->hrgnUpdate || (pWnd->flags & WIN_INTERNAL_PAINT)))) + while (pWnd) { + if (!(pWnd->dwExStyle & WS_EX_TRANSPARENT) && + (pWnd->hrgnUpdate || (pWnd->flags & WIN_INTERNAL_PAINT)) && + GetWindowThreadProcessId( pWnd->hwndSelf, NULL ) == GetCurrentThreadId()) + { + hwndRet = pWnd->hwndSelf; + WIN_ReleaseWndPtr(pWnd); + break; + } WIN_UpdateWndPtr(&pWnd,pWnd->next); - } - if (pWnd) - { - hwndRet = pWnd->hwndSelf; - WIN_ReleaseWndPtr(pWnd); } TRACE("found %04x\n",hwndRet); return hwndRet; -- Alexandre Julliard julliard(a)winehq.com

2 1

Re: Bugs involving returning strings from API functions
by Alexandre Julliard Sept. 7, 2001

Sept. 7, 2001

"Justin Santa Barbara" <justinsb(a)hotmail.com> writes: > As I see it, the advantage of a helper function is that for about the same > amount of work of fixing GetSystemDirectory and GetCurrentDirectory, all the > functions should work the same way meaning any bugs would quickly be found. > It might also be possible to optimize the case where buflen=0, while this > isn't practical per-function. Should I offer a patch only for wine/files/* > ? I'm afraid there are too many variations to make this useful. It would be nice if all APIs that return strings behaved the same way, but unfortunately this is not the case: some count the terminating null, some don't, some copy partial strings when the buffer is too small, some don't copy anything, etc. We would need about as many helpers as there are API functions. Also be aware that the MSDN doc cannot be trusted at all on this matter, the correct behavior needs to be verified for each function. -- Alexandre Julliard julliard(a)winehq.com

1 0

Re: Make cmdline writable (cmdline4)
by Alexandre Julliard Sept. 7, 2001

Sept. 7, 2001

Francois Gouget <fgouget(a)free.fr> writes: > No I don't have a 'real world' exemple where this is a problem. I > became aware of the issue because of warnings when I was working on > dlls/msvcrt/process.c: system is const char* so I made msvcrt_spawn > const char* but then I had to add a cast when calling CreateProcessA > (see second hunk of cmdline3). The cast is wrong, we should do a copy there. CreateProcess takes a char*, not a const char*. Casting off the const and relying on CreateProcess to not modify the string is broken. Even better would be to convert the string to Unicode and call CreateProcessW; this would avoid a copy (well, right now it will cost an extra copy, but that's because CreateProcess should be fixed to do everything in Unicode). -- Alexandre Julliard julliard(a)winehq.com

2 1

Wine for OS X?
by Tyler Vano Sept. 7, 2001

Sept. 7, 2001

Hi All, I'm new to this list (and to Wine itself) but was wondering, is there a project in the works to port Wine to Mac OS X (www.apple.com/macosx)? OS X *is* based on FreeBSD (albeit with a bunch of new fluff for compatibility & aqua), so a port shouldn't be that difficult. (650,000 lines of code and the guy says it 'shouldn't be difficult.') Personally, I think the eventually payoff would be well worth it. You'd get the entire community of ecstatic Mac users, drunk with the power of being able to run Windows apps directly on their Macs (without VPC.) Just like Linux/Unix, Apple's custom Mach3 kernel (aka Darwin) is Open Sourced. Just curious. - Tyler www.creysoft.com __________________________________________________ Do You Yahoo!? Get email alerts & NEW webcam video instant messaging with Yahoo! Messenger http://im.yahoo.com

2 1

Re: Make cmdline writable (cmdline4)
by Alexandre Julliard Sept. 7, 2001

Sept. 7, 2001

Francois Gouget <fgouget(a)free.fr> writes: > CreateProcess may modify the command line that was passed as a > parameter. It does this not in the current process but in the child > process so that the caller may not realize it, but it may hit a snag if > the command line is not writable in the first place. But the command line parameter is a LPSTR, so CreateProcess is allowed to modify it and the caller is not supposed to pass a const string. In what case does this cause a problem? -- Alexandre Julliard julliard(a)winehq.com

2 1

RE: CryptoAPI
by Patrik Stridvall Sept. 6, 2001

Sept. 6, 2001

> > Beside the GPL doesn't cover use, only distribution, so if the user > > would install a Winelib that uses a GPL library that would be > > completely legal as far as that user would be concern despite the > > fact that he has a non-GPL:ed application that uses Winelib. > > Hmm, if only distribution is covered, then can we still write > code for a GPL > library (use) as long as the library isn't shipped with wine > and test for it > on the user's system? Well, actually since the Wine license is GPL compatible that is not a problem either. The problem is that some 3rd party might use Winelib for an application that has non-GPL compatible license. See below. > > However Alexandre has clearly expressed that he doesn't want Wine > > to be a legal test bed, if not primarily for the legal issues, so > > because he doesn't want an unsuspecting company using Winelib be > > "cruficed" on Slashdot or wherever for alleged GPL violation. :-) > > > > While I, in some meaning, agree, I do not believe that we should > > refrain from doing so because of fear of what might never be. > > Obviously we should turn off the use of any GPL:ed library > by default. > > I tend to agree with Alexandre and suggest little or no work > in this area > proceed until the legal debate is over. It seems only best > for the project > to avoid the possibility of legal action against us, > especially for something > we don't fully understand. As far as the Wine source code is concerned the debate as over as it can be without a real court case. The Wine license is GPL compatible so saying that the Wine source code violates the GPL is ridiculous. So what you have to claim is that the Wine source code facillitates others to violate the GPL. This is ridicioulous too for other reasons, foremost because none of requirements of contributary or vicarious (sp?) is even remotely fullfilled. As far as the binaries are concerned, well, the Wine project itself doesn't distribute binaries so it doesn't matter as far as the project is concerned. OK it links to others binaries but claiming liabillity for that is ridiculos too. So far the Wine project. As for other people, well no one really knows since there are very few (if any) relevant court cases on the subject. However consider this, that is actually not a hypotetical situation at all. Even if it is, it still illustrates an example. IIRC when Corel ported Corel Office it used some sort of font rendering library made by Bitstream IIRC. The library cost money to use, but the use of library was optional. If you had it was used if not you didn't get as good fonts as otherwise. So does Corel Office violate Bitstream's copyright? Obviously not, no sane court would claim that and beside Bitstream haven't and won't sue because they benefit from the sitution since now consumers has one more reason to buy their product. Beside that will make it a much harder for them to show damages since their are none. That covers pretty much the use of all normal commersial libraries. In short they have no reason to sue and no chance of success either. Now enter the GPL. The problem is that GPL has an entirely different goal than normal commersial licenses. The commersial library developers wants as many users as possible since each legal user pays money for their product and even illegal users might help them to make the library more widely spread and more of a defacto standard that will likely generate more users (both legal and illegal) in the long run. Note that they normaly don't care what license any user of library distributes his code under either. In short they don't make any difference between different users as long as they are paid. The copyright holder a GPL:ed library on the other hand have no intrest in that software under non-GPL compatible licenses uses their library. This is a fundmental difference. On the other hand the GPL is a distribution not a use license. Paradox? :-) I could speak about this much longer but the point what is legal or what is not is entirely dependent on: 1. Is the GPL is virus that can infect users of the code? 2. If it is a virus, to what extent does the infection spread. Put is another way: GPL:ed libraries discriminates again different kinds of users (developers and end user). The commersial libraries does not normaly do that. Whether and how far copyright law permits this is IMHO one of the most fundamental unresolved questions for open source software development. Even worse. Note that a ruling a _either_ direction have enormous implications for the future of open source software so keep that in mind. "United we stand, divided we fall." Another small note if you are less hostile to the GPL you could say that the GPL is not a virus it is a contract. However all copyright is a contract between the author and society question what contract takes predence is still revelant and unresolved. However if you see the problem with Wine project use of GPL:ed for a contract perspective you could in addition argue that you never have accepted the contract since you didn't have to since you didn't redistribute or modify the library and claim Wine and derived work of the library is rediculous. > > Anyway, unless or until Alexandre changes his mind, I think that > > you should submit all general code that needn't use a specific > > library and distribute that rest as a patch. Preferable so > > that Winelib could dynamically load it so it could be > > distributed seperately in binary form as well. > > Hmm, I never thought about libgcrypt, but if we can somehow > use it legally > then it could be a possible back-end as opposed to or > inconjunction with > OpenSSL. Naturally, I would prefer not to re-invent the > wheel by having to > re-write every needed algorithm used in the M$ API, if we can > find a library > that is legal in all circumstances :-! Use of all normal commersial libraries are "legal" as shown above, at least if you only distribute your Winelib application. Only the GPL and similar viral license can be a problem. > But has anyone considered the following: > -Writing code that interfaces with either OpenSSL or > libgcrypt depending on > what the USER (or any winelib developer) requires AND has > installed on their > system. Providing this is legal. From what I understand > this *should* be > legal long as we're not DISTRIBUTING whatever is illegal to do so. Most probably yes. The only issue as I can see it is how far the GPL extends. > -Use OpenSSL by default if it is more compatible with our own license. > -Disabling the cryptoAPI to winelib developers by default to > make them aware > of these issue and allow them to choose the interface that > matches their > requirements. > -Provide adequate documentation that discusses these points > and touches the > legalities of this software. > > One thing we should probably do no matter what is contact the > developers for > both OpenSSL and libgcrpyt and make them aware of out situation. (and > possibly see which one or both is willing to bend on their > license(s)). That is probably a good idea. > Oh and one other possibility may be to find a way to modify > our own license > such that it fits this problem. Since only GPL libraries AFAICS can be a problem and our license it already GPL compatible that not possible (or needed). > However, I'm not sure how > willing Alexandre > or any of us are about doing this (...again?, I think it's > has been done > before). Well I don't know. Ask him. However he is currently away so that will have to wait I guess. Anyway what you suggested above is fine by me.

1 0

RE: Problem with specmaker
by Patrik Stridvall Sept. 6, 2001

Sept. 6, 2001

> "Guy L. Albertelli" wrote: > > > > Recently had a problem with specmaker not generating > prototypes. Traced it > > down to a problem in function_grep.pl. > > > > If the following lines are in a .h file, no prototypes > after them are > > recognized. Removing the lines from the .h file allows the > prototype to be > > generated: > > > > #define IShellIcon_METHODS \ > > ICOM_METHOD3(HRESULT, GetIconOf, \ > > LPCITEMIDLIST,pidl, UINT,flags, LPINT,lpIconIndex) > > > > I suspect that the preprocessor elimination did not expect multiline > > condition. Well, it knows about multiline directives but it doesn't handle all case properly. > > I don't know perl so I cann't fix it. > thanks for sending the pb to me, but I didn't wrote that part at once > (I just recently upgraded specmaker with some dump facilities) > it seems however that the perl script only allow one backslash for > multiline extensions but fails with more than one > since Patrik seemed to have written this part at once (and it > could/should > be shared with the wineapi suite, I borrowed (cut and paste) code from the winapi suite then I was writing function_grep.pl. It is a crude hack but it works (sort of at least). And yes, function_grep.pl will be part of the winapi suite eventually but probably not in the next few months, but you never know. > I think he's the best suited for the > fix... ;-) Indeed. Here is a working fix I think (not tested). I found a bug in winapi_check a while ago but the fix didn't find its way the function_grep.pl it seems. The problem might also be that there is whitespace after on of the \:es. This is forbidden by ANSI C but most compilers accept it anyway. Hmm. Thinking, modifying etc. What about this? Index: wine/tools/specmaker/function_grep.pl =================================================================== RCS file: /home/wine/wine/tools/specmaker/function_grep.pl,v retrieving revision 1.1 diff -u -u -r1.1 function_grep.pl --- wine/tools/specmaker/function_grep.pl 2001/01/04 19:45:50 1.1 +++ wine/tools/specmaker/function_grep.pl 2001/09/06 21:50:26 @@ -60,11 +60,12 @@ if(/^\s*$/) { next; } # remove preprocessor directives - if(s/^\s*\#/\#/m) { - if(/^\#.*?\\$/m) { + if(s/^\s*\#/\#/s) { + if(/^\#.*?\\\s*\n/s) { $lookahead = 1; next; - } elsif(s/^\#\s*(.*?)(\s+(.*?))?\s*$//m) { + } elsif(s/^\#\s*\w+(?:\s+(.*?))?\s*$//s) { + $again = 1; next; } }

1 0