August 2007 - Wine-Devel - list.winehq.org

Re: [PATCH 2/5] urlmon: Use BSCF_ values passed to report_data to keep track of download state.

by Jacek Caban

Hi Misha, Misha Koshelev wrote: > This simplifies the code a bit by getting rid of an enum and a corresponding variable in the IBinding > implementation, as we can successfully use the bscf values that are passed to report_data to get all > the information. > > Additionally, the original implementation was somewhat incorrect at least for the case of the Http protocol > as far as sending OnStopBinding This patch is already accepted, but unfortunately it's wrong. Tests changed in the attached patch succeed on native urlmon.dll and fail on current Wine. Probably only BSCF_LASTDATANOTIFICATION flag should be used. Other OnProgress callbacks should be called depending on the real binding state, not the one reported by protocol handler. Jacek diff --git a/dlls/urlmon/tests/url.c b/dlls/urlmon/tests/url.c index 187fb03..dcae846 100644 --- a/dlls/urlmon/tests/url.c +++ b/dlls/urlmon/tests/url.c @@ -255,6 +255,7 @@ static HRESULT WINAPI Protocol_Start(IInternetProtocol *iface, LPCWSTR szUrl, SET_EXPECT(OnDataAvailable); SET_EXPECT(OnStopBinding); + bscf = BSCF_LASTDATANOTIFICATION; hres = IInternetProtocolSink_ReportData(pOIProtSink, bscf, 13, 13); ok(hres == S_OK, "ReportData failed: %08x\n", hres);

Aug. 2, 2007

Re: #1 winhttp: Forward WinHttpTime{From, To}SystemTime to their counterparts in wininet.

by Jacek Caban

Hans Leidekker wrote: > We should be able to implement more than 95% of this dll by wrapping/forwarding > to wininet. That's better than many other dlls in Wine and we're already seeing > regressions in apps trying to use our stub winhttp. > And when we'll find an app (I'm sure we will) that uses the remaining 5% of winhttp, we'll have to reimplement it properly, what will be a real pain. I agree that such a big code duplication is ugly, but that's the way to go. We may separate the common code to different files in wininet and keep them in sync with winhttp. This way it shouldn't be too hard to implement most of the functionality you have implemented. Jacek

Aug. 2, 2007

Re: WGL: GetPixelFormat fix for offscreen formats

by Kuba Ober

On Thursday 02 August 2007, Tomas Carnecky wrote: > from the patch: > > return physDev->current_pf; > + TRACE("(%p): returns %d\n", physDev, physDev->current_pf); > } > > What's the purpose of that TRACE() ? Copy/paste bug? Cheers, Kuba

Aug. 2, 2007

Re: WGL: GetPixelFormat fix for offscreen formats

by Tomas Carnecky

from the patch: return physDev->current_pf; + TRACE("(%p): returns %d\n", physDev, physDev->current_pf); } What's the purpose of that TRACE() ? tom

Aug. 2, 2007

Re: #1 winhttp: Forward WinHttpTime{From, To}SystemTime to their counterparts in wininet.

by Hans Leidekker

On Thursday 02 August 2007 20:19:52 Robert Shearman wrote: > We shouldn't be implementing winhttp on top of wininet as there are > functions such as WinHttpSetCredentials > (http://msdn2.microsoft.com/en-us/library/Aa384112.aspx) that can't be > implemented on top of wininet functions. Why not? Could it not add the required headers to the request? > There are also issues with different error codes being returned as well > as different callbacks > (http://msdn2.microsoft.com/en-us/library/aa383917.aspx) Well, error codes can be translated in wrappers and WINHTTP_STATUS_CALLBACK has the same prototype as INTERNET_STATUS_CALLBACK, although it's not called for the same set of events. We should be able to implement more than 95% of this dll by wrapping/forwarding to wininet. That's better than many other dlls in Wine and we're already seeing regressions in apps trying to use our stub winhttp. -Hans

Aug. 2, 2007

Re: #1 winhttp: Forward WinHttpTime{From, To}SystemTime to their counterparts in wininet.

by Robert Shearman

Hans Leidekker wrote: > Forward WinHttpTime{From, To}SystemTime to their counterparts in wininet. > We shouldn't be implementing winhttp on top of wininet as there are functions such as WinHttpSetCredentials (http://msdn2.microsoft.com/en-us/library/Aa384112.aspx) that can't be implemented on top of wininet functions. There are also issues with different error codes being returned as well as different callbacks (http://msdn2.microsoft.com/en-us/library/aa383917.aspx) -- Rob Shearman

Aug. 2, 2007

Re: Is there a process for reviewing a bugzilla staffer?

by Kuba Ober

> >>>> Is there a formal process for reviewing an arguably incompetent > >>>> bugzilla staffer? Obviously it wouldn't be to submit their name as a > >>>> bug. But is there any defined administrative layer that concerns > >>>> itself with people on that level who are dragging on the project? > >>>> > >>>> I looked around a bit for information on this. Sorry if it's posted > >>>> somewhere and I missed it. > >>>> > >>>> Thanks, > >>>> Whit > >>> > >>> To spare everyone time and to skip directly to an entertainment see bug > >>> 9147: http://bugs.winehq.org/show_bug.cgi?id=9147 > >> > >> I agree with Whit. Most of your writing in that bug report would be in > >> line with lack of sleep or prolonged fatigue, or some other factor that > >> causes you to be compartmentalized in your own verions of things and > >> completely ignore the real issue at hand. > >> > >> Similar behavioral pattern that happened on China Airlines Flight 006. > >> The resemblance is striking: the pilots completely unaware of what the > >> real problem is, and were dealing with non-issues (rather than fluing > >> the plane). Similar thing here: the user tells you one thing (the real > >> issue), your situational awareness is as if something entirely different > >> has been taking place (dupes, etc). Interesting. > >> > >> http://en.wikipedia.org/wiki/China_Airlines_Flight_006 > >> > >> http://www.rvs.uni-bielefeld.de/publications/Incidents/DOCS/ComAndRep/Ch > >>inaAir/AAR8603.html > >> > >> http://catless.ncl.ac.uk/Risks/3.79.html > >> > >> [...] the captain had become so preoccupied with the dwindling > >> airspeed that he failed to note that the autopilot, which relied on > >> ailerons only, not the rudder, to maintain heading, was using the > >> maximum left control- wheel deflection available to it to overcome the > >> thrust asymmetry due to the hung outboard engine. When the right wing > >> nevertheless began to drop, ... the captain didn't notice the bank on > >> the attitude indicator ... . When he did notice it, he refused to > >> believe what he saw. At this point, ... the upset had begun and the > >> captain and first officer were both spatially disorientated. > >> > >> You can almost substitute terms from the bugreport for the flying terms > >> above... > >> > >> > >> Common thing to happen when you're tired, distracted, etc. So this is > >> nothing personal, just noticing a pretty common problem. BTDT, one has > >> to learn to recognize the first signs and take a break (helps me). At > >> least here it won't kill anyone. Now, if you *are* sleeping well (and > >> long enough), and are not tired, then IANAD and wouldn't know what to do > >> either... > > I've spent a fair amount of time helping users on irc in #winehq and > > this bug report sounds like one of the most common issues, user error > > precipitated by a distribution that requires a high level of user > > knowledge. The back and forth on the bug is mostly a waste of time > > trying to figure out user compile time options, which version of wine > > is actually running when multiple versions are installed etc. I can > > understand Vitaliy's frustration with this stuff as its easily > > avoidable on almost all binary package based distributions. > > > > Maybe we should point gentoo users at the gentoo wiki page we have and > > enhance that page with things we've learned from gentoo debugging. > As a Gentoo user, I would have to agree. If a fairly good document is > put together a lot of headaches can be avoided. When they are done > following the guide then bugzilla staff and others can help them. The > advantages of this are twofold. One is that there are fewer headaches > from chasing down weird compile time options, etc. The second is that > Gentoo is a more advance/complicated distro and some of the users really > know their stuff and can be quite useful. By providing good > documentation we might encourage those users to participate more. Weed > out the bad, keep the good. Umm, how exactly does Gentoo affect the "legacy" compilation of an autotools-based tarball, where you simply untar, configure and make install which normally will go to /usr/local/...? I'd expect this to work the same on my FC6 system, just like it used to work on a FreeBSD system last time I checked (a few months ago). If wine is not overzealous by design in finding its bits and pieces (I doubt, since I don't recall seeing such a bug) and gets correctly invoked, AFAIK it will ignore any existing installations and do its thing. I'd routinely have version A of wine installed from an rpm, and one or more local versions installed in $HOME for testing before deployment. Somehow it had just worked. Heck, every once in a while when I need two or more versions of wine deployed at the same time from an rpm, that'll work just fine too (a development tool or two used to require an older wine version to work). As a more useful measure, I'd suggest bug reporters who run into possible compilation issues to post their ~/.bash_history with the set of lines starting at untarring the package up to running the test. Should enable easy finger-pointing and take minimum time. In such a case, VM could have said "see line 45 of your history, you've done foo which will not work due to xyz, ask on #wine-users for a correct way of doing it". If there's a better way of providing "steps to reproduce the problem" with a bug report (in this case), I'm all ears. Cheers, Kuba

Aug. 2, 2007

Re: Is there a process for reviewing a bugzilla staffer?

by Kuba Ober

On Thursday 02 August 2007, Chris Morgan wrote: > > Rewriting: > > > > We can conclude that VM was distracted first by the mention of the > > ies4lin and second by attempts to attribute the problem to ies4lin, and > > that, because of these distractions, he was unable to assess properly the > > issue at hand. One also concludes that VM over-relied on the features of > > the development process automation (bugzilla) and this this was also > > causel to tine incident since data available from bugzilla effectively > > masked the real problem. > > I also wanted to mention that ies4lin SHOULD be a distraction. If > users want to use unsupported tools to modify their configuration do > we really want to be debugging those modifications as well as wine > issues? IMHO, you succumb just the same to completely ignoring the real problem at hand. Mind you, flt 006 flight crew was not one person only, yet they were all equally confused (or so it seemed, at least PF and FE were). Nowhere did the bug reporter indicate that he used ies4lin on the failing configuration. "Coincidentally, ies4lin, which runs fine on prior versions of Wine, doesn't accept URL line input in 0.9.42." The user just mentioned this as a coincidence. He then explicitly states: "InfoSelect is broken on a fresh installation before ies4lin is installed to it" Then, VM rambles once more and finally says that he's sorry for missing an attachment. I admit there's a chance that previous installs were not done correctly by WB, just like the #4 engine *did* get hung in flt 006. The deal is that in both cases, the engineer who should have worked the problem did something to exacerbate the issue. VM didn't, IMHO, follow the best approach to get WB to check if he reports the results from a correct install (this is not really documented AFAICS). In 006, the FE didn't follow the correct procedure to un-hang the engine (close bleed air valve). So, now the bug is on the right track. If what Mikolaj Zalewski suggests to check turns out to be the culprit, this will be, in the end, an easy one. Just like flt 006 was an easy one once the pilots went below the clouds and saw what the airplane was really doing. Either way, most of VMs doing was consistent with being generally confused and not letting the bug-reported reality sink in. Again, this is not an indication of incompetence -- well trained people (IIRC human factors) will tend to do it more than untrained ones, given right combination of external factors (fatigue, etc). Heck, being in a grumpy mood can make you act like this -- as long as it's transient, it's something that one can expect to see and can live with :) Cheers, Kuba

Aug. 2, 2007

Re: [1/5] DDraw: Some drivers return DDERR_INVALIDPARAMS if DXTn surfaces aren't supported

by Stefan Dösinger

Am Donnerstag, 2. August 2007 13:57 schrieb H. Verbeet: > On 02/08/07, Stefan Dösinger <stefan(a)codeweavers.com> wrote: > > > If the return value is irrelevant though, what's the point of the test? > > > > I want to catch all the possible return values from windows drivers, even > > if it's just for documentation purposes. > > That doesn't seem very useful if there are no actual applications > depending on those values. The idea is to catch a few things in our test beyond needs of apps known to us. Some of my tests test things even though I had no application that really needed it. So I prefer to document things I come accross, even if I can't see an immediate use, and will most likely never have a use. You never know what the future brings. I do not claim that I have a scheme when writing tests for unused functions or not doing so, it mainly depends on wether I'm in the moot of writing a test :-)

Aug. 2, 2007

Re: Is there a process for reviewing a bugzilla staffer?

by Alex Waite

Chris Morgan wrote: > On 8/2/07, Kuba Ober <kuba(a)mareimbrium.org> wrote: >> On Wednesday 01 August 2007, Vitaliy Margolen wrote: >>> Whit Blauvelt wrote: >>>> Hi, >>>> >>>> Is there a formal process for reviewing an arguably incompetent bugzilla >>>> staffer? Obviously it wouldn't be to submit their name as a bug. But is >>>> there any defined administrative layer that concerns itself with people >>>> on that level who are dragging on the project? >>>> >>>> I looked around a bit for information on this. Sorry if it's posted >>>> somewhere and I missed it. >>>> >>>> Thanks, >>>> Whit >>> To spare everyone time and to skip directly to an entertainment see bug >>> 9147: http://bugs.winehq.org/show_bug.cgi?id=9147 >> I agree with Whit. Most of your writing in that bug report would be in line >> with lack of sleep or prolonged fatigue, or some other factor that causes you >> to be compartmentalized in your own verions of things and completely ignore >> the real issue at hand. >> >> Similar behavioral pattern that happened on China Airlines Flight 006. The >> resemblance is striking: the pilots completely unaware of what the real >> problem is, and were dealing with non-issues (rather than fluing the plane). >> Similar thing here: the user tells you one thing (the real issue), your >> situational awareness is as if something entirely different has been taking >> place (dupes, etc). Interesting. >> >> http://en.wikipedia.org/wiki/China_Airlines_Flight_006 >> >> http://www.rvs.uni-bielefeld.de/publications/Incidents/DOCS/ComAndRep/China… >> >> http://catless.ncl.ac.uk/Risks/3.79.html >> >> [...] the captain had become so preoccupied with the dwindling >> airspeed that he failed to note that the autopilot, which relied on ailerons >> only, not the rudder, to maintain heading, was using the maximum left control- >> wheel deflection available to it to overcome the thrust asymmetry due to the >> hung outboard engine. When the right wing nevertheless began to drop, ... >> the captain didn't notice the bank on the attitude indicator ... . When he >> did notice it, he refused to believe what he saw. At this point, ... the >> upset had begun and the captain and first officer were both spatially >> disorientated. >> >> You can almost substitute terms from the bugreport for the flying terms >> above... >> >> >> Common thing to happen when you're tired, distracted, etc. So this is nothing >> personal, just noticing a pretty common problem. BTDT, one has to learn to >> recognize the first signs and take a break (helps me). At least here it won't >> kill anyone. Now, if you *are* sleeping well (and long enough), and are not >> tired, then IANAD and wouldn't know what to do either... >> >> Cheers, Kuba > > I've spent a fair amount of time helping users on irc in #winehq and > this bug report sounds like one of the most common issues, user error > precipitated by a distribution that requires a high level of user > knowledge. The back and forth on the bug is mostly a waste of time > trying to figure out user compile time options, which version of wine > is actually running when multiple versions are installed etc. I can > understand Vitaliy's frustration with this stuff as its easily > avoidable on almost all binary package based distributions. > > Maybe we should point gentoo users at the gentoo wiki page we have and > enhance that page with things we've learned from gentoo debugging. As a Gentoo user, I would have to agree. If a fairly good document is put together a lot of headaches can be avoided. When they are done following the guide then bugzilla staff and others can help them. The advantages of this are twofold. One is that there are fewer headaches from chasing down weird compile time options, etc. The second is that Gentoo is a more advance/complicated distro and some of the users really know their stuff and can be quite useful. By providing good documentation we might encourage those users to participate more. Weed out the bad, keep the good. > > Chris ---Alex

Aug. 2, 2007