@zetalyrae Approaching three decades old now, but the SUIF 1.x compiler series had multiple binaries that you could use roughly this way. You could do a compiler pipeline in a shell script. I can't recall if you could do textual stdin/stdout piping of IR though.
A great moment for open standards software: Intel has signed an agreement to acquire Codeplay to accelerate multi-architecture, multi-platform implementations of SYCL and oneAPI. https://t.co/4t0K8WRbsC
The Khronos Group Releases OpenCL 3.0 Provisional Specifications.
Learn how OpenCL 3.0 realigns the OpenCL roadmap to enable developer-requested functionality and flexibility.
https://t.co/K5wc1bFJpN
@Neuralsandwich@7thsanctum I think the dishes are meant to be shared but people have taken them off of the spinning thing and are treating them as one dish each. Leaving the exasperated looking person hungry.
@nagyegrimate@VulkanAPI Mantle was a strong influence on Vulkan, but Vulkan had the benefit of even more people refining the design and building its ecosystem. https://t.co/4ZlHX95ZC8
@TartanLlama Works for me: https://t.co/FNyR2CDvF8
Works in Clang. An ICE with GCC, but by definition the fact that GCC can't handle it is a compiler error.
@cjdb_ns And for textbooks etc I think (but haven't measured) that CC style licenses are much more common, but I don't happen to know of any specific advantages that caused that.
@cjdb_ns My point probably doesn't matter for that case, as I assume the code is unlikely to be copied directly into another codebase but rather principles learnt will be applied.
@cjdb_ns For software? Apache 2.0 is more common for software and thus within a company it is more likely to have already been vetted. Might (or might not) make it easier for a developer to integrate at work.