-Phil.
> On May 27, 2016, at 2:23 PM, Alexandr Scherbatiy > <alexandr.scherba...@oracle.com> wrote: > > The following article claims that DPI virtualization of not DPI–aware > applications is available from Windows Vista: > https://msdn.microsoft.com/en-us/library/windows/desktop/dn469266%28v=vs.85%29.aspx > > I just tried to set dpiAware=false in the java.manifest file on my Windows 7 > and the whole Java application was scaled. > That sounds wrong since we still have the api call. > It looks like that for platforms from Windows Windows 8 and higher the Where do you read this comes into effect from Windows 8 ? > fix JDK-8076545 can be reverted because the Windows OS can automatically > scale an application. > For Windows XP the fix JDK-8076545 needs to be reverted to take the > personalized windows font into account. > .. And xp .. > I run SwingSet2 on JDK 8u with the dpiAware=false option on Windows 7 with > scale 150% and Windows 8.1 with scale 200% and found nothing wrong with it. > What does "nothing wrong mean" ? It seems to contradict what you just said about w7 auto scaling your app. > As I understand, the fix JDK-6829055 Update application manifests with new > Windows 7 dpiAware section > was just an improvement. > > Should I remove the dpiAware option from the java.manifiest file in the > current fix as well? I already said we should not do that. There may be surprises like fx not calling the api so now no one does and you will regress that. Phil > > Thanks, > Alexandr. > > >> On 5/27/2016 9:32 PM, Phil Race wrote: >> But my point is that on Windows 8.1 the "125%" font setting is no longer a >> font setting as such. >> Either the *whole* app gets scaled, or *nothing* get scaled, so the user >> complaining about >> this bug would not have seen his request honoured in the same way as is on >> Windows 7. >> >> Or is setting dpiaware causing us to continue to get "XP Style" scaling as >> it is known. >> >> Disabling that manifest entry and reverting to using API is something we >> should also be considering >> but I don't know that this bug is the place to address that change which may >> cause more problems >> as well as fix some .. >> >> -phil. >> >>> On 05/27/2016 11:14 AM, Alexandr Scherbatiy wrote: >>>> On 5/27/2016 8:14 PM, Phil Race wrote: >>>> >>>> From what I read on the web Windows Vista, 7, and 8, Windows uses this >>>> setting to scale just >>>> the fonts, but for 8.1 and 10 it is a whole application scale, so perhaps >>>> rather than just >>>> revert the fix, you can make it OS version dependent ? >>> >>> To allow the Windows OS scale a java application we need to revert the fix >>> which sets dpiAvare=true in the java.manifest file. See JDK-8080153 Cannot >>> disable DPI awareness. >>> >>> Thanks, >>> Alexandr. >>>> >>>> Surface Pro devices are not going to be running Vista or 7 .. and I expect >>>> most are running >>>> Windows 8.1 or later by now. >>>> >>>> >>>> Of course you will need to find a way to test this hypothesis : probably >>>> on a windows 10 >>>> desktop with 125% scaling to see what happens with the Win L&F. >>>> >>>> -phil. >>>> >>>> >>>>> On 05/27/2016 09:44 AM, Alexandr Scherbatiy wrote: >>>>> >>>>> Hello, >>>>> >>>>> Could you review the fix: >>>>> bug: https://bugs.openjdk.java.net/browse/JDK-8157838 >>>>> webrev: http://cr.openjdk.java.net/~alexsch/8157838/webrev.00 >>>>> >>>>> This is the regression from the fix JDK-8076545 in the JDK 8u. >>>>> There was a request JDK-8152980 to backport the fix to JDK 8u because a >>>>> text size is very big on Windows 10 with UI scaling set to 300% when the >>>>> Windows L&F is used. >>>>> >>>>> However, this leads to another issue that a personalized Windows font >>>>> size is not taken into account in Swing app with Windows L&F on JDK 8u. >>>>> >>>>> JDK 9 does not run into this issue because it contains support for HiDPI >>>>> displays which reads the system DPI and scales a whole application with >>>>> fonts according to system UI scale. >>>>> >>>>> The proposed solution is just to revert back the initial JDK-8076545 fix >>>>> for JDK 8u only. >>>>> >>>>> Thanks, >>>>> Alexandr. >
