On 12/03/2012 01:53 PM, Joshua Cranmer wrote:> I don't know any hard examples off the top of my head, but I do > definitely remember (while grepping through python docs earlier today) > being surprised that some of the functions I use on a consistent basis > turned out to have a minimum of python 2.6.I have nothing concrete to point to from the LLVM world, but from my own experience with my own scripting I found maintaining compatibility between Python 2.5 and 3.0 to be a tedious and time consuming process. I suspect there's much more of a maintenance burden here than might be first obvious. A second point worth making is that currently Python is not a dependence of someone building or installing LLVM/Clang. (Er, that hasn't changed recently right?) It's only a dependence for those running the tests (i.e. developers). Not sure how that changes the discussion if at all, but I thought it was worth mentioning. Yours, Philip Reames
> -----Original Message----- > From: llvmdev-bounces at cs.uiuc.edu [mailto:llvmdev-bounces at cs.uiuc.edu] > On Behalf Of Philip Reames > Sent: Monday, December 03, 2012 9:51 PM > To: llvmdev at cs.uiuc.edu > Subject: Re: [LLVMdev] Minimum Python Version > > > > On 12/03/2012 01:53 PM, Joshua Cranmer wrote: > > I don't know any hard examples off the top of my head, but I do > > definitely remember (while grepping through python docs earlier today) > > being surprised that some of the functions I use on a consistent basis > > turned out to have a minimum of python 2.6. > I have nothing concrete to point to from the LLVM world, but from my own > experience with my own scripting I found maintaining compatibility between > Python 2.5 and 3.0 to be a tedious and time consuming process. > I suspect there's much more of a maintenance burden here than might be > first obvious. > > A second point worth making is that currently Python is not a dependence of > someone building or installing LLVM/Clang. (Er, that hasn't changed recently > right?) It's only a dependence for those running the tests (i.e. developers). > Not sure how that changes the discussion if at all, but I thought it was worth > mentioning. > > Yours, > Philip ReamesSlightly OT, Last I checked (yesterday), Python has become a requirement for configuring with CMake, with a base LLVM (no clang, compiler-rt, etc) on Windows, with all tests / examples / tools disabled and only building llvm-core and X86/ARM targets. If anyone knows whether python is actually needed here or just an artifact of the build system, it would be helpful. -Gordon
Gordon Keiser <gkeiser at arxan.com> writes:> Slightly OT, > > Last I checked (yesterday), Python has become a requirement for > configuring with CMake, with a base LLVM (no clang, compiler-rt, etc) > on Windows, with all tests / examples / tools disabled and only > building llvm-core and X86/ARM targets.Unless it changed *very* recently, LLVM tools are built by the default target (`all') on Windows (both VS and Mingw). Tests should work out of the box too. Examples are not built by default, but you can either add them to the `all' target (with -DLLVM_BUILD_EXAMPLES=ON) or explictly build them using its associated targets. By default, only X86 (not ARM) is built on Windows, but you can change that. See LLVM_TARGETS_TO_BUILD in http://www.llvm.org/docs/CMake.html Since yesterday, you twisted your memory quite a bit :-)> If anyone knows whether python is actually needed here or just an > artifact of the build system, it would be helpful.Yes, Python is required by the LLVM CMake build. Possibly by the configure&make build too (I'm not sure, but the plan was for that build to depend on Python too.) CMake doesn't *need* Python at all, the LLVM CMake build worked without Python for years.