July 2009 - Wine-Devel - list.winehq.org

Re: [1/3] WineD3D: An indirect address op can adjust min and max at the same time
by Henri Verbeet July 8, 2009

July 8, 2009

2009/7/7 Stefan Dösinger <stefan(a)codeweavers.com>: > if (reg->idx <= ((IWineD3DVertexShaderImpl *)This)->min_rel_offset) > + { > ((IWineD3DVertexShaderImpl *)This)->min_rel_offset = reg->idx; > - else if (reg->idx >= ((IWineD3DVertexShaderImpl *)This)->max_rel_offset) > + } > + if (reg->idx >= ((IWineD3DVertexShaderImpl *)This)->max_rel_offset) > + { > ((IWineD3DVertexShaderImpl *)This)->max_rel_offset = reg->idx; > + } If you're going to fix it, you might as well fix it properly. There's no point in assigning min_rel_offset or max_rel_offset if they're already the same as reg->idx.

2 1

AcceptEx Patch Comments
by Mike Kaplinskiy July 8, 2009

July 8, 2009

I'm looking for some comments regarding my (winsock part by Scott) acceptex patch: http://bugs2.winehq.org/attachment.cgi?id=22193 . This is just preliminary and doesn't support receive (yet). The approach was done to fix the server-side issues AJ had with Scott's original patch, so server side code would be my main concern right now. Thanks, Mike.

1 0

Re: [PATCH 2/3] jscript: Changed to_primitive implementation
by Jacek Caban July 8, 2009

July 8, 2009

Hi Piotr, Piotr Caban wrote: > --- > dlls/jscript/date.c | 2 +- > dlls/jscript/dispex.c | 2 +- > dlls/jscript/engine.c | 12 ++++---- > dlls/jscript/jscript.h | 9 ++++++- > dlls/jscript/jsutils.c | 57 > ++++++++++++++++++++++++++++++++++++++++---- > dlls/jscript/tests/api.js | 2 + > dlls/jscript/tests/lang.js | 7 +++++ > 7 files changed, 77 insertions(+), 14 deletions(-) > > + hres = jsdisp_get_id(jsdisp, hint == HINT_STRING ? toStringW : valueOfW, 0, &id); + if(SUCCEEDED(hres)) { + hres = jsdisp_call(jsdisp, id, ctx->lcid, DISPATCH_METHOD, &dp, ret, ei, NULL /*FIXME*/); + if(FAILED(hres)) { + FIXME("throw TypeError\n"); + return hres; You leak jsdisp here. + } + else if(V_VT(ret) != VT_DISPATCH) { + jsdisp_release(jsdisp); + return S_OK; + } + } You leak V_DISPATCH(ret) here. + hres = jsdisp_get_id(jsdisp, hint == HINT_STRING ? valueOfW : toStringW, 0, &id); + if(SUCCEEDED(hres)) { + hres = jsdisp_call(jsdisp, id, ctx->lcid, DISPATCH_METHOD, &dp, ret, ei, NULL /*FIXME*/); + if(FAILED(hres)) { + FIXME("throw TypeError\n"); + return hres; + } You leak jsdisp here. + else if(V_VT(ret) != VT_DISPATCH) { + jsdisp_release(jsdisp); + return S_OK; + } + } +date = new Date(); +function toString() { + return "toString"; +} +date.toString = toString; +ok(""+date === "toString", "\"\"+date = " + date); It would be nice to add a few more tests for different cases like returning objects from toString implementation. Also adding toString function to global scope is asking for a trouble in future tests, so I'd suggest construct like date.toString = function() {...} Thanks, Jacek

1 0

Re: user32/tests: Added test for infinite WM_PAINT loop (resend)
by Austin Lund July 7, 2009

July 7, 2009

2009/7/5 Austin Lund <austin.lund(a)gmail.com>: > The last time I sent this, the patch was broken. I've tested it on > WinXPsp2 and wine-1.1.25. This patch relates to bug 10522. > Can I please have some feedback on this patch? The test is for what I found as the source of the problem, not for the loop itself. Should I be testing for the loop instead? Is there an easy way of doing this? For further rambling on the topic see the bug report (http://bugs.winehq.org/show_bug.cgi?id=10522)

1 0

Re: winhttp(0/5): add preliminary IPv6 support
by Austin English July 7, 2009

July 7, 2009

On Tue, Jul 7, 2009 at 4:06 PM, Juan Lang<juan.lang(a)gmail.com> wrote: > Hi, this small patch series adds preliminary IPv6 support to winhttp. > I say it's preliminary because 1) it's not well tested, and 2) it > doesn't attempt to parse IPv6-specific address components such as the > scope ID. Mainly I wanted to get rid of the assumption that every > host address is represented by a struct sockaddr_in, because the > underlying library functions are more general than that. > > I'd appreciate feedback. For instance, it may well be that this > doesn't build on all systems, and that more configure magic will be > necessary. Builds fine for me on Ubuntu Jaunty and PC-BSD 7.1. OpenSolaris builds fine, but adds some warnings about visibility attribute (these are all over the tree though). I don't currently have Net/OpenBSD installed to test on, so if they're broken, I'll bug you when I test next ;-). -- -Austin

1 0

transl and sublangs are not showing neutral
by Ricardo Filipe July 7, 2009

July 7, 2009

hi mikolaj, in transl the Portuguese language (not Portugese :( ) is not showing most of the neutral strings for it's sublangs (portugal and brazil). i think this is not how it should be... in each sublang it should show the Neutral derived strings and the warning that neutral is being used and should not be used aside the language name in the main list, not put neutral separate in another hidden page. as it is now it's hard for me to know what is translated and what's not. what do you (and wine devel) think? could you please change this? regards, Ricardo

1 0

Re: gdi32: Use a modified version of MulDiv to match native behaviour.
by Paul Vriens July 7, 2009

July 7, 2009

Huw Davies wrote: > --- > dlls/gdi32/metafile.c | 46 ++++++++++++++++++------- > dlls/gdi32/tests/metafile.c | 79 ++++++++++++++++++++++++++++++++----------- > 2 files changed, 92 insertions(+), 33 deletions(-) > > > ------------------------------------------------------------------------ > > Hi Huw, You may have already seen this. The changed tests introduce a great number of test failures on Win9x/WinME. I also see 2 test failures on my NT4+ boxes (except Vista) but not for others on test.winehq.org. I had a quick look and this is again the result of my display size being 997/568 (Vmware). As soon as I go full screen these 2 tests succeed. Could you have a look? If you want me to test something just shout. -- Cheers, Paul.

1 0

Fwd: Windows 2D GDI benchmark tools?
by Luke Benstead July 7, 2009

July 7, 2009

---------- Forwarded message ---------- From: Luke Benstead <kazade(a)gmail.com> Date: 2009/7/7 Subject: Re: Windows 2D GDI benchmark tools? To: Roderick Colenbrander <thunderbird2k(a)gmail.com> 2009/7/7 Roderick Colenbrander <thunderbird2k(a)gmail.com>: > Hi, > > I'm working on rewriting Wine its blitting code using XRender. This > has several advantages e.g. no need for depth conversion, no need for > manual stretching/mirroring, it saves a lot of back-and-forth copying > between X and further various operations will be accelerated by the > GPU using XRender. (For a big part it might eliminate the need for a > DIB engine!) > > Right now I'm looking for some 2D GDI benchmark tools which test > blitting performance and other 2D operations and which work on Wine. I > have tried Sisoft Sandra and PassMark but those don't work on Wine. > Does anyone know any other 2D benchmark tools? It would help me a lot > to isolate slow parts and optimize my code even further. > > Thanks, > Roderick > > > (To everyone this time - why am I incapable of hitting "Reply to all" ? ) Maybe this? http://crystalmark.info/software/CrystalMark/index-e.html ( I haven't tried it, just Googled for benchmarking tools - particularly see this screenshot: http://crystalmark.info/software/CrystalMark/images/CrystalMark2004R2_Resul… there is a GDI mark) Luke.

1 1

Try #3: user32: Windows test request (cursors/icons)
by Daniel Santos July 7, 2009

July 7, 2009

Ok, I hope this one is stable on win9x. Please give it a try and let me know. Thanks Daniel

2 1

Re: winex11.drv: Handle clipboard on an auxiliary thread for windowless apps
by Alexandre Julliard July 7, 2009

July 7, 2009

Yuri Khan <yurivkhan(a)gmail.com> writes: > @@ -239,6 +239,9 @@ static Bool filter_event( Display *display, XEvent *event, char *arg ) > case PropertyNotify: > case ClientMessage: > return (mask & QS_POSTMESSAGE) != 0; > + case SelectionClear: > + case SelectionRequest: > + return 1; > default: > return (mask & QS_SENDMESSAGE) != 0; There's no reason to change this, QS_SENDMESSAGE is perfectly appropriate for selection events. -- Alexandre Julliard julliard(a)winehq.org

1 0

Re: msi: Start a dummy thread to initialize an MTA.
by Alexandre Julliard July 7, 2009

July 7, 2009

Hans Leidekker <hans(a)codeweavers.com> writes: > Based on a patch by Dan Kegel. Fixes Adobe CS3/CS4 installers. > See http://bugs.winehq.org/show_bug.cgi?id=18070 That's ugly, please let's fix this the right way, we have all the necessary infrastructure now. -- Alexandre Julliard julliard(a)winehq.org

1 0

What's with winetest-latest?
by Aurimas Fišeras July 7, 2009

July 7, 2009

I get this Error: "Can't enumerate test files: 1813". And from test.winehq.org looks like everyone is getting an error.

3 3

Can resources and translations match Windows?
by Paul Vriens July 7, 2009

July 7, 2009

Hi, This question came to me when I saw Austin's patch about adjusting notepad resources to more closely match XP. Is there some kind of copyright on the resources? Are we allowed to use the same strings or should it always be our own words (like with comments in the source)? -- Cheers, Paul.

3 3

Windows 2D GDI benchmark tools?
by Roderick Colenbrander July 7, 2009

July 7, 2009

Hi, I'm working on rewriting Wine its blitting code using XRender. This has several advantages e.g. no need for depth conversion, no need for manual stretching/mirroring, it saves a lot of back-and-forth copying between X and further various operations will be accelerated by the GPU using XRender. (For a big part it might eliminate the need for a DIB engine!) Right now I'm looking for some 2D GDI benchmark tools which test blitting performance and other 2D operations and which work on Wine. I have tried Sisoft Sandra and PassMark but those don't work on Wine. Does anyone know any other 2D benchmark tools? It would help me a lot to isolate slow parts and optimize my code even further. Thanks, Roderick

1 0

Build a6beb5066884 w2k-sp4-fr-netwrkdrv test results
by Nicolas Le Cam July 7, 2009

July 7, 2009

Hello, Strangely the network drive has been disconnected during the test. Could you remove it ? I will rerun and resubmit it after checking that results are correct. -- Nicolas Le Cam

1 0

Re: [2/3] WineD3D: Make context_attach_surface_fbo fbo aware
by Henri Verbeet July 6, 2009

July 6, 2009

2009/7/6 Stefan Dösinger <stefan(a)codeweavers.com>: > Subject: [PATCH] WineD3D: Make context_attach_surface_fbo fbo aware I hope you meant "sRGB" there. > context_apply_attachment_filter_states((IWineD3DDevice *)This, surface, TRUE); > > - GL_EXTCALL(glFramebufferTexture2DEXT(fbo_target, GL_COLOR_ATTACHMENT0_EXT + idx, surface_impl->glDescription.target, > - surface_impl->glDescription.textureName, surface_impl->glDescription.level)); > + switch(srgb) > + { > + case SRGB_BOTH: > + ERR("SRGB_BOTH fbo attachment requested\n"); > + /* drop through */ > + case SRGB_SRGB: > + GL_EXTCALL(glFramebufferTexture2DEXT(fbo_target, GL_COLOR_ATTACHMENT0_EXT > + idx, surface_impl->glDescription.target, > + surface_impl->glDescription.srgbTextureName, surface_impl->glDescription.level)); > + break; > + > + case SRGB_ANY: /* TODO: Keep track of the last use */ > + case SRGB_RGB: > + GL_EXTCALL(glFramebufferTexture2DEXT(fbo_target, GL_COLOR_ATTACHMENT0_EXT > + idx, surface_impl->glDescription.target, > + surface_impl->glDescription.textureName, surface_impl->glDescription.level)); > + break; > + } I don't think SRGB_BOTH and SRGB_ANY make sense here. I also don't think this can work like this, context_apply_attachment_filter_states() needs to be aware of which texture was requested.

2 1

Re: [3/3] WineD3D: Update sRGB copies with fbo_blit if possible
by Henri Verbeet July 6, 2009

July 6, 2009

2009/7/6 Stefan Dösinger <stefan(a)codeweavers.com>: > +void fbo_srgb_load(IWineD3DDevice *iface, IWineD3DSurface *surface, BOOL dst_srgb) > +{ > + IWineD3DDeviceImpl *This = (IWineD3DDeviceImpl *) iface; > + IWineD3DSurfaceImpl *surf = (IWineD3DSurfaceImpl *) surface; > + > + ENTER_GL(); This code needs an active GL context.

2 1

Re: crypt32: Update german translation (depends)
by Paul Vriens July 6, 2009

July 6, 2009

André Hentschel wrote: > This is for the case Paul Vriens patch gets in > (http://www.winehq.org/pipermail/wine-patches/2009-July/075450.html) > --- > dlls/crypt32/crypt32_De.rc | 1 + > 1 files changed, 1 insertions(+), 0 deletions(-) > > > > > ------------------------------------------------------------------------ > > Hi André, There is no dependency here. I merely changed one of the English resources, I didn't add one. -- Cheers, Paul.

2 2

Re: Try #2: user32: Windows test request (cursors/icons)
by Daniel Santos July 6, 2009

July 6, 2009

> The way you are checking for Win9x is discouraged. We > rather not have GetVersion() but instead rely on behaviour > to check for Win9x. The checks I've inserted for Win9x are strictly to prevent crashing the OS. There were some previous checks in there and since I've had to add more checks, I made an inline function for it. Prior to NT, ms wasn't checking validity of handles in many cases, so calls to SetCursor and DestroyCursor with known invalid handles are what I believe is bringing them down. All of those should now be disabled on win9x. I wouldn't normally focus on such a thing so intensively (passing invalid handles), but this is precisely the type of issue that's been causing hangs and other annoyances in Lord of the Rings Online since their upgrade late last year, which started me working on these fixes. > test_DestroyCursorIcon returns a BOOL but most of the calls > don't use the returned value. A few do but then the return > value is not checked against, for example: > 996 /* Clean up. */ > 997 ret = > test_DestroyCursorIcon(handle, TRUE, TRUE, __LINE__, > 0xdeadbeef); Yea, when I added test_DestroyCursorIcon, I replaced all of the calls to DestroyCursor and DestroyIcon that didn't have special checks like ok(!ret, broken(ret), "..."), but I wanted the return value to be available in case it was needed later in the test code. I've removed all of the cases where the return value was captured in a variable and then never used and now nothing uses the return value, but I figure it's probably better to leave it returning the value than remove it in case some test modification later needs it for something. > There are also some spelling fixes needed (minor though, > I'd guess): hehe, lots of spelling errors, thanks! > 22 * FIXME: Add tests for CreateCursor and verify > that width & hight cannot exceed > Btw. This FIXME doesn't look right. The problem is specifically with the CreateCursor function. When I analysed it, I couldn't find any place that max height & width were checked and the GetSystemMetrics calls for those values are just hard-coded to return 32. Personally, I don't understand the X Windows implications well enough to delve further at this point, but I know that we are definitely missing tests for CreateCursor as well as the hidden 16-bit API function DestroyIcon32() which I've modified in my patches (yet to be submitted). > 78 * shouldFail [I] Rather of not it should > fail > Not sure what you are trying to say here. I clarified this documentation better. Essentially, TRUE if the call to DestroyCursor or DestroyIcon is expected to fail. I think I'll have the time so I'm going to try to get a new patch to you today. Thanks! Daniel

1 0

Re: Try #2: user32: Windows test request (cursors/icons)
by Daniel Santos July 6, 2009

July 6, 2009

Thank you again for your time, effort and feedback! I wont have much time today, but I'll get another rendition as soon as I can. Daniel --- On Mon, 7/6/09, Paul Vriens <paul.vriens.wine(a)gmail.com> wrote: > From: Paul Vriens <paul.vriens.wine(a)gmail.com> > Subject: Re: Try #2: user32: Windows test request (cursors/icons) > To: daniel.santos(a)pobox.com > Cc: wine-devel(a)winehq.org > Date: Monday, July 6, 2009, 3:49 AM > Daniel Santos wrote: > > Thanks again for testing this for me, Paul. I > have modified the tests so that they should pass and I've > skipped tests on win9x that I suspect are causing the blue > screens (personally, I doubt it's due to VMware). > > > > I moved the new test functions to the top of the file, > but I didn't remove the comments. I can shorten them > if preferred, but I personally prefer to have functions well > documented. > > > > So if you or anybody else could test this again for > me, I would greatly appreciate it. > > > > I guess it will be AJ's call in the end if these types of > comments are necessary. I don't have a strong opinion about > them. > > The way you are checking for Win9x is discouraged. We > rather not have GetVersion() but instead rely on behaviour > to check for Win9x. > > test_DestroyCursorIcon returns a BOOL but most of the calls > don't use the returned value. A few do but then the return > value is not checked against, for example: > > 996 /* Clean up. */ > 997 ret = > test_DestroyCursorIcon(handle, TRUE, TRUE, __LINE__, > 0xdeadbeef); > > There are also some spelling fixes needed (minor though, > I'd guess): > > (hight -> height) > 22 * FIXME: Add tests for CreateCursor and verify > that width & hight cannot exceed > Btw. This FIXME doesn't look right. > > 78 * shouldFail [I] Rather of not it should > fail > Not sure what you are trying to say here. > > (consistenit -> consistent) > (accross -> across) > 166 * Only use on very simple test cases where > behavior is consistenit accross > > (destory -> destroy) > 1365 /* Try to destory a bunch of > crap */ > FWIW, I don't like the word crap, but that's probably just > me. > > Test results: > > Win95: > cursoricon.c:1003: Test failed: Last error is 3735928559 > (0xdeadbeef), expected ERROR_INVALID_CURSOR_HANDLE (1402). > cursoricon.c:1004: Test failed: Last error is 3735928559 > (0xdeadbeef), expected ERROR_INVALID_CURSOR_HANDLE (1402). > cursoricon.c:305: Test failed: DestroyCursor returned 1, > expected FALSE. > cursoricon.c:305: Test failed: Last error is 3735928559 > (0xdeadbeef), expected 1402 (0x57a). > cursoricon.c:307: Test failed: DestroyIcon returned 1, > expected FALSE. > cursoricon.c:307: Test failed: Last error is 3735928559 > (0xdeadbeef), expected 1402 (0x57a). > cursoricon: 6 tests executed (0 marked as todo, 0 > failures), 0 skipped. > cursoricon: 495 tests executed (0 marked as todo, 6 > failures), 0 skipped. > > Running it a second time hangs the box !!! > > Win98: > > Logged the output to file but then notepad.exe crashed. > Soon after that the system hangs !!!!! > > So Win9x needs definitely some more work. > > NT4: No failures. > > XP: No failures. > > Didn't test W2K, W2K3 and Vista as the output previously > was the same as XP. > > -- Cheers, > > Paul. >

1 0

installshield and 3GB ram (affects Dragon Naturally Speaking)
by Dan Kegel July 6, 2009

July 6, 2009

Jay, does the problem affect any other apps? Until some confirms the patch works and/or that other apps are affected, the bug might not get much attention. - Dan ---------- Forwarded message ---------- From: Jay Starkman <jay(a)starkman.com> Date: Mon, Jul 6, 2009 at 5:38 AM Subject: Bug 15708 To: bugs-admin There's a patch posted for Bug 15708, LARGE_ADDRESS_AWARE flag. In attempting to install Dragon Naturally Speaking 7.0, all of us have gotten the message "DNS requires...at least 128MB RAM. Your system has...-673MB RAM." The setup program sees negative RAM on systems with over 2G. Since the patch is already out there, is it possible to compile the next iteration WITH the patch? I tried compiling my own WINE using the patch, but it caused problems and the WINE I compiled didn't work at all. I don't see Bug 15708 in your list of Bugs Scheduled to be Fixed before WINE 2.0 is released. There appears to be a simple patch that fixes it. Wouldn't it be possible to incorporate that patch into your next Wine_1.1.26? -Jay Starkman -------------------------------------------------------------------------- Jay Starkman, CPA Voice: 404-636-1400 Fax: 404-636-1130 2531 Briarcliff Road, Suite 116 e-mail: j(a)starkman.com Atlanta, Georgia 30329 Internet: http://www.starkman.com --------------------------------------------------------------------------

1 0

Re: crypt32: Add "Disallowed" to CryptFindLocalizedName
by Paul Vriens July 6, 2009

July 6, 2009

Aurimas Fišeras wrote: > "Disallowed" is Used in Chrome. > > > ------------------------------------------------------------------------ > > Hi Aurimas, I was trying to update the NL resources but wasn't sure what to translate to. Disallowed on an English Windows version is actually "Untrusted Certificates". I will sent a "fix" for the English resources as the above makes more sense then 'Disallowed', I think. -- Cheers, Paul.

1 0

Re: why is Kronenberg's Wine/Mac work blacklisted on winehq?
by James Mckenzie July 6, 2009

July 6, 2009

>James McKenzie wrote: >>Something that I do have to point out is that freetype, a very >important >>library for Wine is not shipped with XCode nor MacOSX. We have to >>include it. > >Huh? On my 10.5.7 Leopard system, Wine was able to find FreeType -- with >LD_LIBRARY_PATH guidance. And I have nothing but >- MacOS (with iLife preinstalled) >- (selected packages from) the Xcode install DVD and >- XQuartz 2.3.3.2 >installed on that box. > >Did you install Apple's X11 before installing XQuartz, as the XQuartz >site recommends? > >Are you mislead by the following error message? Possibly. I will check my system and see if libfreetype exists in /usr/X11/lib. James McKenzie

1 0

Re: Direct3D 10 Current Status
by Henri Verbeet July 6, 2009

July 6, 2009

Chris Smowton wrote: > Hello, > > First of all my apologies if you're the wrong person to ask; I got your > name by searching the wine-cvs list for who seemed to be doing most > D3D10 work. If you're not the right person to ask the following > questions, I'd appreciate if you'd advise who if anyone I'd be better > off asking. > In general, wine-devel is the most appropriate place for questions like these. > Now then, I'd like to ask you about the current state of Wine's D3D10 > support. I'm looking at how much effort will be required to gain D3D10 > support for virtual machines, and Wine's D3D9 code has of course been > used in a few virtualisation solutions. Here are my specific questions: > > 1. Would I be right in assuming that D3D10 support is currently very > basic -- at the level of user apps being able to create a device and > context, but not perform rendering as of yet? > Pretty much. We can do some rendering, but don't expect anything beyond very simple sample applications to work. > 2. What about shaders -- has the Wine project made much progress > supporting the new shader models? > We have some initial SM4 support. We can reuse a lot of the infrastructure for previous shader models, but the SM4 binary format isn't publicly documented anywhere, as far as I'm aware. > 3. Whilst of course Wine doesn't need to do this, have you gained any > insight into how the MS D3D10 usermode runtime interfaces with the DWM? > I assume that simply replacing d3d10.dll on a Windows system with the > hope of remoting rendering is not practicable because the runtime > presumably creates the window's primary surface with a shared handle, > and passes that handle on to the DWM to perform rendering. I see no > reason why that internal D3D usermode runtime <--> window manager > process interaction ought to be documented, and so a replacement for > d3d10.dll on a real Windows system would seem like it would necessarily > involve reverse engineering that process, but perhaps I'm wrong? > We didn't really look. I'd expect that to be done by dxgi rather than d3d10 or d3d10core though. Henri

1 0

[rfc] lstrcmpi: order still wrong (was "Re: Regression in lstrcmpiA (occurred in late June, NLS related)" from 2003 year)
by Yuriy Kaminskiy July 6, 2009

July 6, 2009

Hello! Previous thread on this topic: http://www.mail-archive.com/wine-devel(a)winehq.org/msg01080.html I've stumbled over problem with lstrcmpi sorting is still wrong. Some japanese game engine uses binary search on presorted array, and fails with a-la "object not found" errors. Judging by object order in archive, === cut === ... conf_p.MGD----- (would fail with strcasecmp, ok with wine) conf01.MGD--/ ... title.MGD------ fails with vanilla wine title_p.MGD--/ ... === cut === proper order should be "_" < "0" (ok) and "." < "_" (fails with vanilla wine). I've replaced collation weight of '_' with 0x02560111, and now these games run fine; but that's dirty hack, of cause, and should not be applied to upstream: 1) it is modifies generated file; 2) weight for "_" chosen arbitrary and can cause conflicts somewhere else (or, rather, not can, but certainly will - there are other symbols with weight 0x0256???); 3) weight for other "_"-like chars should be modified too. Hope you can suggest better solution. FWIW, I've checked mentioned in previous thread unicode-2.1.9d8 tables - same mismatch, will not work too. I think, only proper way is somehow extract this table from windows (either directly by LCMapStringW(LC_MAP_SORTKEY), or sorting array of a[i]=i; with CompareStringW and using that order). I'm not a lawyer, but really doubt that such reproduced table can be considered copyrightable anywhere. How can anyone make compatible reimplementation without reproducing in some way this table? --

2 3