As you can see from our support table, Windows support is in a sad state these days. There are both proximate and root causes for this.
The proximate cause is issue #540. LLVM 12 removed the ORC MCJit wrapper, and while (ironically) LLVM continues to support the original MCJit, this does not work on Windows. The specific error can be seen in that issue, but likely this requires migration to ORCv2 #485. (The good news is that we waited so long to do this transition that we can skip ORCv1 entirely :joy:.) That is just a hunch though, and someone would actually need to look into this to be sure.
The root cause is that no one is putting effort into Windows specifically. I don't have time to put any nontrivial effort into it, nor do I even have a Windows dev box (or any intention to get one). I can keep the lights on via CI, but this is not a sustainable situation. I also frankly just don't have patience for the peculiarities of the platform. So, if our Windows support is going to stay, it would really be best for it to be maintained by someone who actually cares, has Windows (and ideally LLVM) experience, and can put effort into this over the long term.
If anyone wants to get started they can reply here or one of the linked issues, or discuss it on Zulip.
Hi all,
As you can see from our support table, Windows support is in a sad state these days. There are both proximate and root causes for this.
The proximate cause is issue #540. LLVM 12 removed the ORC MCJit wrapper, and while (ironically) LLVM continues to support the original MCJit, this does not work on Windows. The specific error can be seen in that issue, but likely this requires migration to ORCv2 #485. (The good news is that we waited so long to do this transition that we can skip ORCv1 entirely :joy:.) That is just a hunch though, and someone would actually need to look into this to be sure.
The root cause is that no one is putting effort into Windows specifically. I don't have time to put any nontrivial effort into it, nor do I even have a Windows dev box (or any intention to get one). I can keep the lights on via CI, but this is not a sustainable situation. I also frankly just don't have patience for the peculiarities of the platform. So, if our Windows support is going to stay, it would really be best for it to be maintained by someone who actually cares, has Windows (and ideally LLVM) experience, and can put effort into this over the long term.
If anyone wants to get started they can reply here or one of the linked issues, or discuss it on Zulip.
Thanks!