- Introduce a helper function to query the console leader process (instead of using duplicated code).
- Fix the algorithm of the last close notification, as demonstrated by the tests I did on windows 2003 (test app is provided in CORE-7250):
a console app that registered for the last close notification, closes only if the process that was the console leader process at the time the app registered for the notification, is killed.
In this case, we notify the app, and we clear some flags. On the contrary, if we close the app that registered for the notification, we just clear the flags without doing extra operations.
svn path=/trunk/; revision=62861
- formatting
- remove BOM from the Russian resource
- update the Czech translation
- add myself to the credits because vanity~
svn path=/trunk/; revision=62857
Do not ignore failure when calling DrvEnablePDEV and properly handle a failure in PDEVOBJ_bEnablePDEV instead of ASSERTing. Properly cleanup the PDEV after failure, check if the PDEV was enabled before calling DrvDisablePDEV. Fixes some crashes with VBox driver.
svn path=/trunk/; revision=62855
- Add a debug output to the default console dispatcher for the CTRL_LAST_CLOSE_EVENT code;
- hardcoded_value--;
Addendum to revision 62847.
svn path=/trunk/; revision=62848
Implement (undocumented) SetLastConsoleEventActive API.
Patch by Alexander Andrejevic, with a minor modification by me concerning the addition of a NotifiedLastCloseProcess member to the CONSOLE structure and a check that makes sure that only the app that asked for the notification, receives it (and so that we don't call the console control dispatcher for nothing).
This API is used by ntvdm to be sure that it gets killed when all other console apps attached to the ntvdm's console are away.
CORE-7250
svn path=/trunk/; revision=62847
Make kernel32 / winsrv console CSR structures Win2k3-compliant for CreateConsoleScreenBuffer.
The aim is to be able to put our kernel32.dll or winsrv.dll on win2k3, and vice-versa.
Because of that, we need to introduce and use the redundant hMutex and lpBitMap in the CONSOLE_CREATESCREENBUFFER structure, instead of the ones that are already present in its CONSOLE_GRAPHICS_BUFFER_INFO member ...
Isn't MS dumb sometimes?
I also homogeneize some variable names wrt. equivalent ones that I use elsewhere, and I fix a broken return value in the case the API fails.
Part 3/X
CORE-7931
svn path=/trunk/; revision=62846
