On 2017-12-17 Carsten Haitzler <ras...@rasterman.com> wrote: > On Sun, 17 Dec 2017 07:21:41 +0100 Andreas Metzler <ametz...@bebt.de> said: [...] >> luajit generally has support for arm64 but EFL breaks with "bad light >> userdata pointer" (T6156), so as a workaround we have switched to >> lua-old on Debian/arm64.
> well luajit didn't even document this limitation until fairly recently > (end of 2016). in fact it's a limitation for x86-64 too... for a long > time before being documented. it's a very very very odd thing to limit > and no one would expect a pointer can only use a limited number of > bits for the address unless clearly and distinctly documented. (the > pull request came in end of 2016, and luajit is STILL broken for 52bit > addressing for example). my take is that luajit is basically broken on > 64bit as it makes major assumptions on the VALUE of an address - if > the kernel were to remap elsewhere in virtual memory in certain ways > it'd still be broken if we just used basic raw pointers. adding the > documentation at least kind of helps... but it's still a major design > error. we've worked around it in git by now. Hello, my layman's reading of the respective bug-reports also was that the respective luajit interface was broken. Is the suggested solution ("switch to FFI bindings") a real solution? Will the workaround in EFL GIT possibly end up in a 1.20.x release or is t strictly 1.21 material? TIA, cu Andreas -- `What a good friend you are to him, Dr. Maturin. His other friends are so grateful to you.' `I sew his ears on from time to time, sure' ------------------------------------------------------------------------------ Check out the vibrant tech community on one of the world's most engaging tech sites, Slashdot.org! http://sdm.link/slashdot _______________________________________________ enlightenment-devel mailing list enlightenment-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/enlightenment-devel