the point was to use implicit casts instead of thousands of overloads
the point was to use implicit casts instead of thousands of overloads
Mwhat if we break their kneecaps
Pi guess the handle scenario needs to be resolved regardless if we want minimal unsafe in a codebase
Mshould GLFWwindow have methods on it?
Pno it's an opaque type
Pit may do eventually
Mwhat does the handle do, besides hiding the T*
Pas we do want to experiment with autogenerated OOP wrappers
P(or rather me and kai)
Pnothing
Pthat is all it exists for
Mso it's still unsafe
Pyes and this is what i always unsuccessfully try to communicate to our users
Mmaybe the customer isn't always right
Pit doesn't stop people coming in here time and time again whining about the pointers for having the gall to exist and infiltrate their lovely safe codebase
Mever tried asking them for an alternative design?
Pno they're users
Pkai tells me off when i explode on users for not contributing
Mi don't mean as a genuine question, i mean as a way to systematically destroy their argument
M"what would you rather have instead of a pointer here"
Pthey will and i shit you not say
IntPtrPdon't test them
Mif GLFWwindow is opaque anyway, instead of something like StoredPtr<GLFWwindow>, just have a type called GLFWwindowPtr. i think IMGUI.NET does that
Phmmm yeah that's not bad
Pugh i'm so tired, i don't want to be working on 3.0 i only restarted out of a sense of obligation, can someone please take over
DI have been experimenting with a source generator to do this. Works ok with WebGPU bindings. Have not been game try it with vulkan because I am too lazy to write a test app. 
Nalmost caught myself cloning silk recursively
L40gb
Nyep. good thing i did Not
Aas in internet bandwidth or..?
because my internet is shared it can sometimes die without warning but that's only happened like twice now
otherwise it's actually pretty fast
because my internet is shared it can sometimes die without warning but that's only happened like twice now
otherwise it's actually pretty fast
Await bandwid--- ah whatever i think i get what you mean
not internet bandwidth like my tired ass this morning thought
yeah i dont mind helping out, just let me know what you're after
not internet bandwidth like my tired ass this morning thought
yeah i dont mind helping out, just let me know what you're after
Aenglish is not my strong point
A.. as is evident lol
Mi'm writing scriban templates to make generating the new silk.net.maths types easier
Pbasically SilkTouchX is kinda easy to work with, so I was wondering if someone could try coming up with an alternative overload mechanism
Asure i don't mind taking a look
Pawesome, if you need any help roslyn quoter is your friend
Aalrighty
Pi think my main focus for now is going to be getting the bindings generated
Psomeone else can figure out the overload situation as it's depressing me
Aby overloads i assume you do mean like
the function overloads
and not something else i'm missing right?
the function overloads
and not something else i'm missing right?
Aand you're looking for a mechanism to generate good overloads without generating billions of them?
Pbasically the whole "using functions 'safely'" problem, so function overloads and user storage of pointers are the main things.
Pand yeah would not like 50+ overloads for one function
Agotcha
Ai have ideas for opengl functions, not sure how well it would transfer over to vulkan etc
Abut its all worth experimenting with
Pyeah exactly
Pthe SilkTouchX pipeline is super easy to hook into with the mod system, just make a mod, add it to the mod loader, add it to your config and you immediately start receiving all the syntax
PGLFW is the test bench for now but I'll have OpenGL online in hopefully not long
Aalright awesome
