This reverts commit 0.4.8-dev-502-g
7000fe2340.
This reverts commit 0.4.8-dev-503-g
f318a25e10.
The two changes were done once to work towards better
shim-supporting APIs of newer Windows versions on demand
in the future. Luckily we do not depend on them yet.
Giannis already fixed most regressions of the 2 commits
before branching 0.4.8RC (e.g. 2nd stage theme-selector).
Unfortunately we still have issues atm with the existing mixture
of Wine comctl32 v5 vs v6.
The image-lists of both are not compatible as Giannis told us.
We revert both commits for this release to fix CORE-14433
(missing icons due to incompatible image-lists). We can not
simply add a manifest here, because 3rd party apps are affected.
I double-checked, all of those (formerly affected) apps
are working fine and show the icons after the revert:
2nd-stage-theme-selection, MS Excel Viewer 2007 open-dlg,
MSVCPP 6 installer open-dlg, FaxViewer save-dlg,
MS Word Viewer 2003 open-dlg
This revert also obsoletes works/fixes like
2f11904000.
To prevent the visible symptom
'sporadic NTOSKRNL work queue deadlock when exiting command.com'
This partially reverts r58615 / 01729482ae.
(cherry picked from commit 3159de5573)
cherry-picked by Joachim Henze
To prevent the visible symptom
'sporadic NTOSKRNL work queue deadlock when exiting command.com'
(cherry picked from commit 2b8f4f6e24)
cherry-picked by Joachim Henze
This commit was *not* committed to master yet.
I'll leave the ticket unresolved until it'll be.
This fixes a regression introduced in SVN r73442:
Our setup created undesired temporary files and left them
in the temp folder. Also we saw a slight one-time-increase
in memory consumption.
This patch was the work of Carlo Bramini
He proposed the fix on 2016-12-30 already.
There are ongoing discussions about an alternative approach
to the initial problem that SVN r73442 tried to address.
Serge Gautherie can and shall be evaluate that alternative
later during 0.4.9dev.
For now many thanks to Carlo Bramini for this fix.
This commit was *not* committed to master yet.
I'll leave the ticket unresolved until it'll be.
We had the palette for some games regressed starting with r75061
when using VBEMP (and our inbuilt Mesa, not the rapps Mesa).
This wasn't noticed early because the same commit led to
many games crashing as well. After we recovered from the crashes
later, we were confronted with a regressed palette state.
E.g. in Diablo II, Anno 1602, MS Age of Empires.
With that commit the palette now looks like in r75060 again.
Palette in Rapps Mesa is now equal again to palette inbuilt Mesa.
The palette is still not perfect -
(CORE-10738 still needs to be addressed later).
Many thanks to zefklop!
A bit of history: in r71528, I tried to fix our explorer often
crashing while browsing directories. It was linked to the fact
that a notification result may arrive while the notification
structure had already been deleted.
The fix for this was actually broken and was leading to a double
leak: the notification structure was leaked. But also the handle
to the directory that had been browsed!
This means that the directory couldn't be modified anymore as
a leaked handle to it was still open.
Actually, when notifications are cancel, the kernel properly
calls the notification routine, but with a specific error code.
So the correct fix is to stop handling that notification when
we receive this error code. This is the correct fix with no leaks.
This commit is a complete r71528 revert with the appropriate fix.
CORE-10941
CORE-12843
(cherry picked from commit da8a41b97b)
cherry-picked by Joachim Henze with many thanks to Pierre.
This shows that we may have problems in the future with applications using just comctl32 v5
(cherry picked from commit 2f11904000)
cherry-picked by Joachim Henze
This avoids locking Cc for too long by trying to read-ahead data which
is already in cache.
We now will only schedule a read ahead if next read should bring us
to a new VACB (perhaps not in cache).
This notably fixes Inkscape setup which was slown down by read-ahead
due to continous 1 byte reads.
Thanks to Thomas for his help on this issue.
CORE-14395
(cherry picked from commit c4f58bbfd8)
cherry-picked by Joachim Henze
This reverts bfcab8b0b4
I was sceptical regarding that patch already when I committed it.
I revert it now because it would introduce more regressions like
CORE-14354. The risk is not worth it.
Ofc this will bring back BSOD CORE-13907 & CORE-11678.
- Ask the user if they want to continue on cab certificate problems
- Fix memory leaks in failure paths
- Properly disable caching
(cherry picked from commit 07f3691cd8)
cherry-picked by Joachim Henze
- We should not open a new connection to request a certificate.
- Update the issuer / subject for the LE certificate.
- User proper types.
- Require all fields that we check to be present.
Without checking the public key or the entire certificate it's still not secure, but we are a step closer.
Dedicated to Joachim Henze
CORE-14350
cherry-picked by Joachim Henze
object2.patch in CORE-11678 is from jimtabor and Work in Progress.
As expected it mitigates CORE-11678 by preventing the
sporadic BSOD upon closing CCleaner. But it will always lead to
the program remaining in taskmgr upon closing instead.
User can manually kill the app from taskmgr.
To my own surprise this patch also reliably hides the BSOD in
HxD portable for me (CORE-13907).
I am still sceptical regarding this patch and will carefully
watch out for surprises like zombie processes during the test-cycle.
Pierre recommended this hack-fix for 0.4.8rls before.
Avoids "GetVolumeInformation now fails on NFS volume"
I will not retest any RDBSS functionality myself in my tests,
maybe some other tester would be kind enough to retest this?
Please note that I replaced #if 0 with #if 1
as discussed with Pierre. That's different to the patch in ticket.
Thomas linked the patch in Wine Bug 44410
I tested it for some weeks without noticing any side-effects.
It avoids text rendering regressions in some setups like UltraISO
This patch reduces failures of SHGetFileInfo function. CORE-7159
* improve WideByteToWideChar calls
* fix bugs related to file attributes and SHGFI_EXETYPE
* SHGFI_USEFILEATTRIBUTES and SHGFI_ICON fix
* s/sizeof(temppsfi)/0/
CDefView: Make GetItemObject return an IContectMenu after setting itself as the site.
CDefViewBckgrndMenu: Make the inner context menu use the same site. Implement handling CMDSTR_VIEWLISTA and CMDSTR_VIEWDETAILSA.
CDefaultContextMenu: CMDSTR_VIEWLISTA and CMDSTR_VIEWDETAILSA shouldn't be handled in this class.
Fixes CORE-14327 and the two buttons that change the view mode
It shouldn't be necessary for QueryContextMenu to be called before InvokeCommand.
Move retrieving the IContextMenu of the folder to the initialization of the class.
CORE-12866
