May 2017 - Wine-Devel - list.winehq.org

Re: [1/5] wsdapi: Implement WSD LinkedMemory functions [try 4]

by Owen Rudge

On 02/05/2017 21:49, Alexandre Julliard wrote: > Are all allocations supposed to be zeroed? This is a little surprising, > it would need a test too. Good point - it seems not. I can't think how I could write a test to reliably prove that the allocation shouldn't be zeroed (as occasionally by chance it will be), so I presume I should be OK to just amend the implementation to remove the zeroing. > Please don't do that, test results need to be reproducible. Sorry, I'll submit an amended test to fix that. Thanks, -- Owen Rudge http://www.owenrudge.net/

May 3, 2017

Re: [RFC] README: Update Linux version requirement to 2.3.31.

by Alexandre Julliard

Bruno Jesus <00cpxxx(a)gmail.com> writes: > On Wed, May 3, 2017 at 4:25 PM, <wylda(a)volny.cz> wrote: >>> >>>> I've not tested the version, but recent commit 45a632ee81c9c5c58309426de5aadbd8dbbba0f5 >>>> introduces SYS_mmap2 system call which doesn't exist before 2.3.31. >>>> So, should we update the requirement? Is 2.3.31 valid enough? >>> >>>I'm not sure that we know what the minimum version is these days. If >>>mmap2 is really the only new dependency, adding a workaround is easy >>>enough. Is anybody able to test this? >> >> Hmmm, no answers... It seems to me like crazy project, which nobody on Earth >> would ever undergo. To promote it from crazy to totally insane I could give a drive >> to one of my oldies. Like... > > Isn't commit https://source.winehq.org/git/wine.git/commitdiff/9f06a197c8c156acf21d8d809… > about this? For that specific issue, yes. But it would be interesting to find out if there are other issues that we aren't aware of, and whether they would be worth trying to fix. Of course it's low priority, so do it only if that's your idea of fun... -- Alexandre Julliard julliard(a)winehq.org

May 3, 2017

Re: [RFC] README: Update Linux version requirement to 2.3.31.

by Alexandre Julliard

<wylda(a)volny.cz> writes: > So when testing old kernels (Debian distributions) – If that ever compile what > else could be interesting? GCC version? Libc version? X version? > > How to interpret successful test > - compilation? (pre-compiled in VM) > - Running winecfg?? > - Running MS Paint??? If you can ever get it to compile, running winecfg is probably good enough as a test. If you run into trouble (as I expect), knowing which version of which component is causing trouble would be interesting. -- Alexandre Julliard julliard(a)winehq.org

May 3, 2017

Re: [RFC] README: Update Linux version requirement to 2.3.31.

by Bruno Jesus

On Wed, May 3, 2017 at 4:25 PM, <wylda(a)volny.cz> wrote: >> >>> I've not tested the version, but recent commit 45a632ee81c9c5c58309426de5aadbd8dbbba0f5 >>> introduces SYS_mmap2 system call which doesn't exist before 2.3.31. >>> So, should we update the requirement? Is 2.3.31 valid enough? >> >>I'm not sure that we know what the minimum version is these days. If >>mmap2 is really the only new dependency, adding a workaround is easy >>enough. Is anybody able to test this? > > Hmmm, no answers... It seems to me like crazy project, which nobody on Earth > would ever undergo. To promote it from crazy to totally insane I could give a drive > to one of my oldies. Like... Isn't commit https://source.winehq.org/git/wine.git/commitdiff/9f06a197c8c156acf21d8d809… about this?

May 3, 2017

Re: [RFC] README: Update Linux version requirement to 2.3.31.

by wylda＠volny.cz

> >> I've not tested the version, but recent commit 45a632ee81c9c5c58309426de5aadbd8dbbba0f5 >> introduces SYS_mmap2 system call which doesn't exist before 2.3.31. >> So, should we update the requirement? Is 2.3.31 valid enough? > >I'm not sure that we know what the minimum version is these days. If >mmap2 is really the only new dependency, adding a workaround is easy >enough. Is anybody able to test this? > >-- >Alexandre Julliard >julliard(a)winehq.org > Hmmm, no answers... It seems to me like crazy project, which nobody on Earth would ever undergo. To promote it from crazy to totally insane I could give a drive to one of my oldies. Like - 486DX - 386DX + 387 / 32MB RAM - 386DX / 4MB RAM with ISA Trident 512KB (no 386SX / 486SX;) just to see, how deep we can sink with those old ships :-)) I would let Pentium and Pentium MMX docked for being too modern. So when testing old kernels (Debian distributions) – If that ever compile what else could be interesting? GCC version? Libc version? X version? How to interpret successful test - compilation? (pre-compiled in VM) - Running winecfg?? - Running MS Paint???

May 3, 2017

Re: [PATCH resend] comctl32: When adding a scrollbar to a listview, update the other one.

by Nikolay Sivov

On 03.05.2017 7:53, Alex Henrie wrote: > Cc: Nikolay Sivov <nsivov(a)codeweavers.com> > > Fixes https://bugs.winehq.org/show_bug.cgi?id=35529 > > Adding a vertical scrollbar can necessitate adding a horizontal > scrollbar just as easily as adding a horizontal scrollbar can > necessitate adding a vertical scrollbar. (And in the same way, removing > either of the scrollbars can necessitate removing the other scrollbar.) > > This patch was previously assigned to Nikolay, but I never received any > feedback about it. Sorry about that. > > Signed-off-by: Alex Henrie <alexhenrie24(a)gmail.com> > --- > dlls/comctl32/listview.c | 78 +++++++++++++++++++++++++++++++----------------- > 1 file changed, 50 insertions(+), 28 deletions(-) > > diff --git a/dlls/comctl32/listview.c b/dlls/comctl32/listview.c > index c01d816c46..c3c89d08dc 100644 > --- a/dlls/comctl32/listview.c > +++ b/dlls/comctl32/listview.c > @@ -2016,23 +2016,10 @@ static void LISTVIEW_UpdateHeaderSize(const LISTVIEW_INFO *infoPtr, INT nNewScro > SWP_NOZORDER | SWP_NOACTIVATE); > } > > -/*** > - * DESCRIPTION: > - * Update the scrollbars. This functions should be called whenever > - * the content, size or view changes. > - * > - * PARAMETER(S): > - * [I] infoPtr : valid pointer to the listview structure > - * > - * RETURN: > - * None > - */ > -static void LISTVIEW_UpdateScroll(const LISTVIEW_INFO *infoPtr) > +static INT LISTVIEW_UpdateHScroll(LISTVIEW_INFO *infoPtr) > { > - SCROLLINFO horzInfo, vertInfo; > - INT dx, dy; > - > - if ((infoPtr->dwStyle & LVS_NOSCROLL) || !is_redrawing(infoPtr)) return; > + SCROLLINFO horzInfo; > + INT dx; > > ZeroMemory(&horzInfo, sizeof(SCROLLINFO)); > horzInfo.cbSize = sizeof(SCROLLINFO); > @@ -2084,10 +2071,22 @@ static void LISTVIEW_UpdateScroll(const LISTVIEW_INFO *infoPtr) > dx -= SetScrollInfo(infoPtr->hwndSelf, SB_HORZ, &horzInfo, TRUE); > TRACE("horzInfo=%s\n", debugscrollinfo(&horzInfo)); > > - /* Setting the horizontal scroll can change the listview size > - * (and potentially everything else) so we need to recompute > - * everything again for the vertical scroll > - */ > + /* Update the Header Control */ > + if (infoPtr->hwndHeader) > + { > + horzInfo.fMask = SIF_POS; > + GetScrollInfo(infoPtr->hwndSelf, SB_HORZ, &horzInfo); > + LISTVIEW_UpdateHeaderSize(infoPtr, horzInfo.nPos); > + } > + > + LISTVIEW_UpdateSize(infoPtr); > + return dx; > +} > + > +static INT LISTVIEW_UpdateVScroll(LISTVIEW_INFO *infoPtr) > +{ > + SCROLLINFO vertInfo; > + INT dy; > > ZeroMemory(&vertInfo, sizeof(SCROLLINFO)); > vertInfo.cbSize = sizeof(SCROLLINFO); > @@ -2117,6 +2116,37 @@ static void LISTVIEW_UpdateScroll(const LISTVIEW_INFO *infoPtr) > dy -= SetScrollInfo(infoPtr->hwndSelf, SB_VERT, &vertInfo, TRUE); > TRACE("vertInfo=%s\n", debugscrollinfo(&vertInfo)); > > + LISTVIEW_UpdateSize(infoPtr); > + return dy; > +} > + > +/*** > + * DESCRIPTION: > + * Update the scrollbars. This function should be called whenever > + * the content, size or view changes. > + * > + * PARAMETER(S): > + * [I] infoPtr : valid pointer to the listview structure > + * > + * RETURN: > + * None > + */ > +static void LISTVIEW_UpdateScroll(LISTVIEW_INFO *infoPtr) > +{ > + INT dx, dy; > + > + if ((infoPtr->dwStyle & LVS_NOSCROLL) || !is_redrawing(infoPtr)) return; > + > + /* Setting the horizontal scroll can change the listview size > + * (and potentially everything else) so we need to recompute > + * everything again for the vertical scroll and vice-versa > + */ > + > + dx = LISTVIEW_UpdateHScroll(infoPtr); > + dy = LISTVIEW_UpdateVScroll(infoPtr); > + dx += LISTVIEW_UpdateHScroll(infoPtr); > + dy += LISTVIEW_UpdateVScroll(infoPtr); > + This certainly doesn't look pretty, but I see what you mean. I'll take a look, thanks for reminding me. > /* Change of the range may have changed the scroll pos. If so move the content */ > if (dx != 0 || dy != 0) > { > @@ -2125,14 +2155,6 @@ static void LISTVIEW_UpdateScroll(const LISTVIEW_INFO *infoPtr) > ScrollWindowEx(infoPtr->hwndSelf, dx, dy, &listRect, &listRect, 0, 0, > SW_ERASE | SW_INVALIDATE); > } > - > - /* Update the Header Control */ > - if (infoPtr->hwndHeader) > - { > - horzInfo.fMask = SIF_POS; > - GetScrollInfo(infoPtr->hwndSelf, SB_HORZ, &horzInfo); > - LISTVIEW_UpdateHeaderSize(infoPtr, horzInfo.nPos); > - } > } > > >

May 3, 2017

Re: [PATCH 01/10] d3dx9/tests: Use compare_float() in D3DXFresnelTest().

by Alexandre Julliard

Matteo Bruni <mbruni(a)codeweavers.com> writes: > Signed-off-by: Matteo Bruni <mbruni(a)codeweavers.com> Could you please try to send the signoff mail as a reply to the original patch, instead of a reply to your previous signoff? This way the tracker can find the appropriate patch automatically. Thanks. -- Alexandre Julliard julliard(a)winehq.org

May 3, 2017

Re: [2/4] ole32: Implement DataCache_InitCache

by Huw Davies

On Tue, May 02, 2017 at 06:11:22PM +0800, Wei Xie wrote: > diff --git a/dlls/ole32/datacache.c b/dlls/ole32/datacache.c > index 441a874..51a67e2 100644 > --- a/dlls/ole32/datacache.c > +++ b/dlls/ole32/datacache.c > @@ -2052,8 +2052,45 @@ static HRESULT WINAPI DataCache_InitCache( > IOleCache2* iface, > IDataObject* pDataObject) > { > - FIXME("stub\n"); > - return E_NOTIMPL; > + HRESULT hr; > + FORMATETC fmt; > + STGMEDIUM med; > + > + FIXME("semi-stub\n"); > + > + hr = IDataObject_GetData(pDataObject, &fmt, &med); > + if (FAILED(hr)) > + { > + return CACHE_E_NOCACHE_UPDATED; > + } > + > + if (!fmt.cfFormat || !fmt.tymed || > + (check_valid_clipformat_and_tymed(fmt.cfFormat, fmt.tymed, FALSE) > + && TYMED_NULL != med.tymed)) > + { > + switch(med.tymed) > + { > + case TYMED_GDI: > + { > + fmt.cfFormat = CF_BITMAP; > + fmt.ptd = NULL; > + fmt.dwAspect = DVASPECT_CONTENT; > + fmt.lindex = -1; > + fmt.tymed = TYMED_GDI; > + } > + break; > + default: > + FIXME("Fill formatetc\n"); > + } > + } > + > + hr = IOleCache2_SetData(iface, &fmt, &med, FALSE); > + if (FAILED(hr)) > + { > + return CACHE_E_NOCACHE_UPDATED; > + } > + > + return hr; > } This isn't correct. What I think the tests will show you is that it looks something like the pseudo-code: for_each_cached_format( &fmt ) /* some function that returns the fmt of each cache entry */ { IDataObject_GetData(pDataObject, &fmt, &med); IOleCache2_SetData(iface, &fmt, &med); } Of course there needs to be error handling, tests if the cache has already been initialized, etc. However for now I suggest you just concerntrate on getting the tests in, and come back to this and the subsequent patches later on. Huw.

May 3, 2017

Re: [1/4] ole32/test: Add InitCache test

by Huw Davies

On Tue, May 02, 2017 at 06:10:44PM +0800, Wei Xie wrote: > From 4f8b94b5a8c7739901eae8f19317ca6255911d20 Mon Sep 17 00:00:00 2001 > From: Wei xie <xiewei(a)linuxdeepin.com> > Date: Tue, 2 May 2017 18:07:22 +0800 > Subject: [PATCH 1/4] ole32/test: Add InitCache test ole32/test: -> ole32/tests: > +static HRESULT WINAPI DataObject2_QueryInterface( > + IDataObject* iface, > + REFIID riid, > + void** ppvObject) > +{ > + CHECK_EXPECTED_METHOD("DataObject2_QueryInterface"); > + > + if (IsEqualIID(riid, &IID_IDataObject) || IsEqualIID(riid, &IID_IUnknown)) > + { > + *ppvObject = iface; > + return S_OK; > + } > + *ppvObject = NULL; > + return S_OK; > +} There's no need to create a new IDataObject implementation. Use the existing one and add a file-scope variable to control what _GetData does (which is all you really care about for this test). > +static HRESULT WINAPI DataObject2_GetData( > + IDataObject* iface, > + LPFORMATETC pformatetcIn, > + STGMEDIUM* pmedium) > +{ > + *pformatetcIn = FormatEtc; > + *pmedium = StgMedium; > + return S_OK; > +} You're not checking this method is called, which should be one of the goals of this test! pfomratetcIn is an [in] variable, so assigning things to it is useless. > @@ -1838,10 +1994,23 @@ todo_wine { > > DeleteDC(hdcMem); > > - todo_wine { You remove a todo_wine here and yet this patch doesn't change Wine's implementation, so this will produce a failure. The test must run cleanly on Wine (and Windows) after just this patch is applied. If you fix the implementation later on in the series, then remove the todo_wine in that patch, not this one. > + /* IDataObject_GetData Invalid */ I don't understand how this comment relates to the code. > hr = IOleCache2_InitCache(pOleCache, &DataObject); > ok(hr == CACHE_E_NOCACHE_UPDATED, "IOleCache_InitCache should have returned CACHE_E_NOCACHE_UPDATED instead of 0x%08x\n", hr); > - } > + > + fmtetc.ptd = NULL; > + fmtetc.dwAspect = DVASPECT_CONTENT; > + fmtetc.lindex = -1; > + fmtetc.cfFormat = CF_BITMAP; > + fmtetc.tymed = TYMED_GDI; > + stgmedium.tymed = TYMED_GDI; > + stgmedium.hBitmap = CreateCompatibleBitmap(hdcMem, 100, 100); > + stgmedium.pUnkForRelease = NULL; > + IDataObject_SetData(&DataObject2, &fmtetc, &stgmedium, FALSE); > + > + /* IOleCache2_InitCache not need IOleCache2_Cache first */ > + hr = IOleCache2_InitCache(pOleCache, &DataObject2); > + ok(hr == CACHE_E_NOCACHE_UPDATED, "IOleCache2_InitCache should have returned CACHE_E_NOCACHE_UPDATED instead of 0x%08x\n", hr); Why is the hunk even necessary, you're just calling InitCache again below a test that shows it fails. > > IPersistStorage_Release(pPS); > IViewObject_Release(pViewObject); > @@ -1851,6 +2020,44 @@ todo_wine { > CHECK_NO_EXTRA_METHODS(); > } > > + hdcMem = CreateCompatibleDC(NULL); > + fmtetc.ptd = NULL; > + fmtetc.dwAspect = DVASPECT_CONTENT; > + fmtetc.lindex = -1; > + fmtetc.cfFormat = CF_BITMAP; > + fmtetc.tymed = TYMED_GDI; > + stgmedium.tymed = TYMED_GDI; > + stgmedium.hBitmap = CreateCompatibleBitmap(hdcMem, 100, 100); > + stgmedium.pUnkForRelease = NULL; > + IDataObject_SetData(&DataObject2, &fmtetc, &stgmedium, FALSE); > + > + expected_method_list = methods_cacheinit; > + > + hr = CreateDataCache(NULL, &CLSID_NULL, &IID_IOleCache2, (LPVOID *)&pOleCache); > + ok_ole_success(hr, "CreateDataCache"); > + > + hr = IOleCache2_QueryInterface(pOleCache, &IID_IViewObject2, (LPVOID *)&pViewObject2); > + ok_ole_success(hr, "IOleCache_QueryInterface(IID_IViewObject2)"); > + > + hr = IOleCache2_Cache(pOleCache, &fmtetc, 0, &dwConnection); > + ok_ole_success(hr, "IOleCache2_Cache"); > + hr = IOleCache2_InitCache(pOleCache, &DataObject2); > + ok_ole_success(hr, "IOleCache2_InitCache"); > + > + hr = IViewObject2_SetAdvise(pViewObject2, DVASPECT_CONTENT, ADVF_PRIMEFIRST, &AdviseSink); > + ok_ole_success(hr, "IViewObject2_SetAdvise"); > + > + hr = IViewObject2_GetExtent(pViewObject2, DVASPECT_CONTENT, -1, NULL, &sz ); > + ok_ole_success(hr, "IViewObject2_GetExtent"); > + > + hr = IViewObject2_Draw(pViewObject2, DVASPECT_CONTENT, -1, NULL, NULL, NULL, hdcMem, &rcBounds, NULL, draw_continue, 0xdeadbeef); > + ok_ole_success(hr, "IViewObject2_GetExtent"); > + > + IViewObject2_Release(pViewObject2); > + IOleCache2_Release(pOleCache); > + > + DeleteDC(hdcMem); For this bit of the test, drop the IViewObject calls. What we care about is testing InitCache(). Then I suggest calling _Cache twice with two different formats and checking that _InitCache calls DataObject_GetData for those two formats. You'll need to make DataObject_GetData respond to these two different formats. It's a bit more complicated than that however because in the CF_BITMAP case _InitCache apparently calls _GetData with CF_DIB (this is part of the cache's ability to use dibs instead of bitmaps and vice-versa). You may want to avoid using CF_BITMAP to test this at this stage (we can add CF_BITMAP <-> CF_DIB tests later). Huw.

May 3, 2017

Re: [PATCH resend] comctl32: When adding a scrollbar to a listview, update the other one.

by Alex Henrie

2017-05-02 23:10 GMT-06:00 Alistair Leslie-Hughes <leslie_alistair(a)hotmail.com>: > Hi Alex, > > On 03/05/17 14:53, Alex Henrie wrote: >> Cc: Nikolay Sivov <nsivov(a)codeweavers.com> > >> +static void LISTVIEW_UpdateScroll(LISTVIEW_INFO *infoPtr) >> +{ >> + INT dx, dy; >> + >> + if ((infoPtr->dwStyle & LVS_NOSCROLL) || !is_redrawing(infoPtr)) return; >> + >> + /* Setting the horizontal scroll can change the listview size >> + * (and potentially everything else) so we need to recompute >> + * everything again for the vertical scroll and vice-versa >> + */ >> + >> + dx = LISTVIEW_UpdateHScroll(infoPtr); >> + dy = LISTVIEW_UpdateVScroll(infoPtr); >> + dx += LISTVIEW_UpdateHScroll(infoPtr); >> + dy += LISTVIEW_UpdateVScroll(infoPtr); >> + > Is there any reason your calling update twice for each control? For example: 1. Because of new content, a horizontal scrollbar is added, taking up some of the vertical space (dx = LISTVIEW_UpdateHScroll(infoPtr)) 2. Because there is not enough vertical space now, a vertical scrollbar is added, taking up some of the horizontal space (dy = LISTVIEW_UpdateVScroll(infoPtr)) 3. The horizontal scrollbar thumb's position is updated to reflect that there is less horizontal space now (dx += LISTVIEW_UpdateHScroll(infoPtr)) Or: 1. Because of new content, a vertical scrollbar is added, taking up some of the horizontal space (dy = LISTVIEW_UpdateVScroll(infoPtr)) 2. Because there is not enough horizontal space now, a horizontal scrollbar is added, taking up some of the vertical space (dx += LISTVIEW_UpdateHScroll(infoPtr)) 3. The vertical scrollbar thumb's position is updated to reflect that there is less vertical space now (dy += LISTVIEW_UpdateVScroll(infoPtr)) -Alex

May 3, 2017