cause I'm not sure how well SLN merging works
cause I'm not sure how well SLN merging works
Djust noticed
PIn GitKraken tick both
Psimplest way
PIf you accidentally changed all the other GUIDs though it is more complicated
Pbut if you’ve only added new stuff merge in main and tick both sides on the sln
Dok
TSo apparently we can stop that 'guid constantly changing' thing by specifically setting the <ProjectGuid> field in the csproj's, if it doesnt have one set then apparently the sln's are allowed to just reinvent them as they see fit
PYeah but that’s just so damn ugly
Pand is only a VS thing, whereas we both use Rider so it’s only a concern for incoming contributions
Pwhich are currently few and far between
TThey are, might make it easier to get more though, as we're less likely to see people stick around if using their IDE of choice means they have to manually make sure the guids dont change every time they want to contribute
TEven for just me it currently slows down creating / extending the examples, as adding the csproj's causes the problem 100% of the time
PI wonder whether within common.props we can do something that automatically constructs a GUID based off of the hash of the project name
DWhat is the purpose of SilkTouch generating PtrToString call when you pass a string as param?
DDit marshalls it back
Dthis is something we need to be smarter about eventually. just difficult to know when this is needed and when it is not
Pwe should look at removing that callvirt
Dnot a problem - jit eliminates this
Dthe whole point of the
as GeneratedVTable is this devirtualizationPah ok
Dnow I do wonder what happens when I disable the re-marshalling
Dthat would probably help perf
Pwhen do we remarshal btw?
Dthe whole string marshalling is a bit of a problem anyways, I mean the string marshaller is just a big pile of garbage, it does work, but not perfectly
DI think always? need to check
Pwe only need to remarshal if it’s ref or out string
Ppointless otherwise as strings aren’t passed by ref so that remarshalled value is inaccessible
DI think something like that happens
Dneed to have a look
NHi, Kai. I'm here if you need to ask questions.
DNativeApiGenerator gives CS0208 when using ref string, it's fine with PInvoke
Dyup, there's a bunch of problems around string marshalling like this
D@Perksey please review the OpenAL thing, I want to also fix the bitdepth thing soon
Pwill do
Pjust finishing up on my stuff
Dawesome
DIdk why I always do this, start one week before the update happens, and then my list of things I want to get in grows every day
Phaha
Pi wanted to get more in too
Plike the MSAA thing for OpenGL
Didk why I commented "I'll try making an Avalonia integration for 2.4" like yesterday
Pbut i think i'll put that off until we have a design for better API-specific properties
Dthis is actually impossible lol
Dyup, will wait with the avalonia thing too. 3.0 stuffs
Pmight kick-off 3.0 dev after this
D
Pthere's still a lot of work being put into 2.x, like the freaking resharper plugin
Dyeah, I think we just need to avoid breaking things too hard with 3.0 for now
Dsince we will need merge later
