The StartPrograms processes now all correctly are added
to a list of processes to stop only if the user requested they
are stopped.
Additionally, a 300ms delay was added after SteamGame is
launched to give it time to actually run the darn game! It was
erroring due to testing for the launch too quickly. Steam
takes a wee bit of time to start the game, and we need to
account for that.
The RunShortcut commandline option mostly works
but it fails to cleanly execute thanks to a partially broken
Program.ApplyProfile. The ApplyProfile exception logic
is not working properly and that is causing an exception
in Program.RunShortcut. Just needs some improved
ApplyProfile logic to detect when it does actually
(correctly) fail.
This change loads the start programs into
the form. The structure of the ShortcutItem
changes are designed to be extensible to a
large number of start programs, but the UI
only supports 4. The plan is to add as many
as you want in the WPF version of this app,
but right now its not worth the effort :).
Program to start UI works as intended. It's
a bit basic but will be good enough until
I manage to get to the WPF rewrite in a
year or so! Next step is to start adding the
background changes to the ShortcutItem
to store the information in the JSON.
Had missed the shortcutsave button so it
wold be disabled on load, even if the shortcut
was a valid shortcut. This is now changed so
that it works fine on load.
Have also changed the shortcut library selection
colours to match the rest of the application.
Adding these checkboxes will let the user
easily temporarily disable running a program
without losing the very complicated arguments
passed to the program. This is especially true if
it is a thing like SimHub or something like that!
This is the first design for a simple form to
detail the programs that the user wants us to start
before the main Game. This is a first go and
is only the initial draft.
Made the IsPossible deviceIdentification work
correctly with the NvAPIWrapper logic. Issue
was caused by me not really understanding what
the NVIDIA documentation meant. This has been
rectified. HeliosPlus will now invalidate any profiles
that won't work with the currntly attached screens.
I'd made a revision to the Paths and Topologies
equality comparison and it used Equals for
sequences rather than SequenceEqual as it should.
Comparison works correctly now.
Fixed up some broken logic around detecting
loadedProfiles as the currentProfile. Now when
we load the profiles from disk we also correctly
check whether one of the profiles is in use now
and if it is then we use that profile instead of
creating a new one.
I broke the circularProgressBar but now it works
fine. Messed with the style a bit to get the look
I wanted. Also adjusted the positioning of the
text in the middle.
Added some logic to check if either the from or to
profile conatins an NVIDIA surround screen. If not,
then there is no need to apply a GridTopology as it
is all single devices! Saves up to 15 seconds on
swap over.
Was missing a SetTopologies command from upstream
and the latest GeForce devices don't seem to be able to
read EDID which means the Topologies dont have a
DisplayName peroperty, which in turn causes issues when
trying to change to that topology! So now detect empty
DisplayName and fill it with a default.
Fixed the WindowsDisplayAPI Device Identifcation
logic so that it reliably detects the displays in a way
that works when some displays are disabled as well
as when they are enabled. Also divided up the
DeviceIdentification into two separate functions so
that one will list all attached devices, and the other
will only report the devices currently in use.
DisplayUI is mostly working but it needs some work
fixing the error situations when the PathInfo can't be
applied properly. Need a way of setting the Task
result from the Task Action. Don't know how to do that
so something new to learn!
This code makes the display profile change work. There
are a few things broken still.... namely the showing of a
timer when the profile is being applied, the accuracy of
the display detection code, and the correct logic for
determining a profile can be used.
There are different displayIdentifiers created if the video card
is an NVidia card, or if it is anything else. These identifiers
are saved in the display profiles, and that allows us to compare
the display identifiers we currently have with the ones in various
profiles, and then work out which profiles are valid.
This code committed today ensures that the nvidia device
identifier will always use the NVAPIWrapper lib (whether in
surround mode or not) so that the application allows both
surround mode profiles and non-surround mode profiles
to be selected if the same displays are present.
For all other display adapters it just uses the
WindowsDisplayAPI to create a different displayIdentifier
which isn't as smart. It means that if you have an AMD and
use EyeInfinity, then this isn't clever enough to understand
the profiles use the same matching screens. I will have to
fix this eventually when i build in native AMD support in
the future.
The existing logic for detecting displays connected to
an NVidia device doesn't allow matching for Nvidia
surround and normal surroundless profiles. This means
that you cannot determine which profiles are the valid
for your display setup. This is testing the NVAPIWRapper
scanning and the WindowsDisplayAPI libraries to see what
information will work for tracking the screen in the profiles.
DisplayIdentifiers are the minimum GPU/Display
that will allow us to identify whether a profile
will actually work. We look for the same DisplayIdentifiers
or a subset of them to determine whether a profile
will be valid to show. This is different from the data
stored under DisplayTarget, as that is used for detailed
compatibility....
First part of installing an MSTest suite.
Included some config file to read in
for testing the ShortcutRepository and
the ProfileRepository. Had the data
handy so it made sense to save it :)
The IsPossible logic in the ProfileItem incorrectly
uses the ToPathTargetInfo, when a better match is
to look for all the actual displays that are currently
turned on, and then allow all profiles that include
only the displays that are turned on. That involves
1. getting a list of the display devicepaths, and then
2. checking if this profile has all the devicepaths
needed to show the profile. If it does, then it
IsPossible.
The ProfileAdapter and ShortcutAdapters used
by the ImageListView Control unfortunately
have to access the sizes of the bitmaps being
loaded into the respective imagelistviews. I
can't find anyway of stopping the GDI+
from complaining about the Bitmap being
accessed by multiple different threads (as
ImageListView creates one thread per image.
This will be fixed once I move to this being a
WPF application as we'll use a different control.
Profile selection logic was running functions multiple
times which was resulting in a lot of extra delay
while using the DisplayProfileFOrm. These changes
streamline it and make it work a lot faster!
Moved the shortcut running logic from the
main Program into the ShortcutRepository
so that it can be used from the
Shortcut Library UI as well as from the
commandline.
It's marked as WIP because I also moved
the ApplyingChangesForm and accompanying
ApplyTopos/ApplyProfiles logic and I broke
it. It needs refactoring and hopefully
simplifying but I need to work on that next!
