Josh

I never finish anyth...

[Josh @joshibasa]

@bherbruckx

[Josh @joshibasa]

@TheGingerBill @dkharms_ arent these the default spacings of WinUI? do you mean they could have come up with a more compact, custom widget or something?

[Josh @joshibasa]

@raysan5 @KDCoder agents propose this. user starts developing something with AI and raylib, agent thinks it found a bug in raylib, asks user "should I create a ticket?". it's literally that easy for a user to flood the issue trackers of any public project. harnesses should guard this I think

[Josh @joshibasa]

@SebAaltonen @sulka as a developer I understand you 100%. as a gamer it would frustrate me and no I definitely don't read anything. maybe if you test for these cases during launch, you can show a message with an explanation and redirect to refund page? I think that would make me happy.

[Josh @joshibasa]

@hisownhero @bherbruckx exactly this. I believe there is no other mature kernel other than opencascade and all AI knows about fillets and chamfers comes from that. completely underwhelming. solidcraft is looking nice tho. just needs to take parasolid as an example for solidcore

[Josh @joshibasa]

@sharadsatoskar @syfert @heyshrutimishra what are you on about? it literally means you can have same world knowledge with 1/10 th size model. if this is not very valuable knowledge to you, you should just refrain from commenting weird takes on it

[Josh @joshibasa]

@sharadsatoskar @syfert @heyshrutimishra Is 1 small? How about 0.000000000001? Scale is always relative, you don't need the word "relative" for that to be clear. 109M is smaller than 72B.

[Josh @joshibasa]

@kilua_here @filpizlo you dont need fil-c to do that. it still wont even scratch the surface

[Josh @joshibasa]

@lowkeytekk @ID_AA_Carmack Training data of AI was written by the same humans you deem lazy or unskilled.

[Josh @joshibasa]

@hosseeb Qwen Max and GLM-5.3 are not even close. Qwen wins by an exponential difference.

[Josh @joshibasa]

@anduriltech

[Josh @joshibasa]

@gosdfndskj @nicopreme why? is it too hard to just npm install? you can always install but you can't remove integrated stuff

[Josh @joshibasa]

@Love2Code used oneplus, redmi, xiaomi for years. then I switched to pixel 10 pro. what an awful downgrade. I regret my decision.

[Josh @joshibasa]

@Elijah_Meeks this is pretty much SOTA vector rendering. In theory, AI is guaranteed to *not* generate anything SOTA as it's a slave to its training data.

[Josh @joshibasa]

@forgebitz thats literally why AI generates slop. It was trained on slop

[Josh @joshibasa]

@jonnymevs this must be a ragebait

[Josh @joshibasa]

@film_girl this is purely political. they are losing money. they need regulations so that competition is guaranteed to slow down as well. thus they feed this AGI/sentient agenda

[Josh @joshibasa]

@howard @andruyeung my man.... religion exists because church doesnt want people to think. I really do hope it's a joke otherwise it's time to excersize that brain for you

[Josh @joshibasa]

@nickbernstein @yawaramin @rachelnabors @ryanleichty @dhh you didn't read the article did you

[Josh @joshibasa]

@filpizlo @Love2Code Have you ever tried Qwen-3.8 Max? In xhigh thinking mode? I cant trust other models anymore. This one thinks for hours but finds the correct answer each and everytime for me. more expensive than 5.3-flash though