LLVM's Go Front-End Was Finally Dropped From The Official Source Tree
Most probably didn't even realize LLVM had a Go language front-end, but this past week it was dropped from the official source mono repository.
This LLVM Go front-end "LLGO" hasn't been maintained in several years and never really took off... Most probably aren't even aware of this Go compiler support for LLVM. So the code has been suffering, it was stuck at Go version 1.5 well behind the latest upstream, it likely has build errors, and there are other nuisances with the code like having an entire copy of Mark Twain's "The Adventures of Tom Sawyer" novel. For those wondering why an entire novel was part of the source tree, it amounted to serving as a compression test case.
Should anyone step up to revive LLGO, it can be reverted or there is also external LLGO copies. But at least as far as the official LLVM source tree is concerned, LLGO is removed. However, more than likely any future LLVM Go implementation would involve a complete rewrite as opposed to reviving this old code-base, similar to what we've seen out of the improving Fortran compiler front-end support.
Stripping out the defunct LLGO shortened up the code-base by 729,809 lines.
This LLVM Go front-end "LLGO" hasn't been maintained in several years and never really took off... Most probably aren't even aware of this Go compiler support for LLVM. So the code has been suffering, it was stuck at Go version 1.5 well behind the latest upstream, it likely has build errors, and there are other nuisances with the code like having an entire copy of Mark Twain's "The Adventures of Tom Sawyer" novel. For those wondering why an entire novel was part of the source tree, it amounted to serving as a compression test case.
Should anyone step up to revive LLGO, it can be reverted or there is also external LLGO copies. But at least as far as the official LLVM source tree is concerned, LLGO is removed. However, more than likely any future LLVM Go implementation would involve a complete rewrite as opposed to reviving this old code-base, similar to what we've seen out of the improving Fortran compiler front-end support.
Stripping out the defunct LLGO shortened up the code-base by 729,809 lines.
16 Comments