On Sat, Dec 15, 2012 at 06:41:14PM +0000, Chris Gransden wrote: > In article <20121215171742.gh15...@kyllikki.org>, > Vincent Sanders <vi...@netsurf-browser.org> wrote: > > > Please can you send a patch? even if its just the files you changed I > > can work up a suitable patch from those, whatevers easiest. > > Steve already beat me to it. Those changes are working.
good to know > > > How do we disable JIT? I used your recipie to build spidermonkey as a > > template. Even with --disable-tracejit passed to configure METHOD JIT > > is still enabled (toolchain build has the patches and Makefile used[1] > > (in the sdk directory)). If I turn that off too (--disable-methodjit) > > the library fails to build, perhaps I am missing something? > > > I would like to try and debug where the later version of spidermonkey > > fails as it is such a huge improvement in functionality going > > forwards. > > I think the JIT is built if ARM is detected. Setting > WTF_CPU_ARM_TRADITIONAL 0 in Platform.h and '-DENABLE_ASSEMBLER=0 > -DENABLE_JIT=0' stops it building the JIT. > might try this today though I have run out of time for messing with toolchain stuff right now. > > I can build 1.7 temporarily, but this is not really solving the issue > > at all and will (from past experience) almost certianly lead to RISC > > OS being stuck there forever. > > > In future I do wish to move onto editions of spidermonkey past 1.8.5 > > (3 years old) which provide improved perfomance and functionality so > > limiting myself to version 1.7 (which is over five years old now, think > > firefox 2 era) seems counterproductive. > > I thought it would be useful for getting people to test stuff now. I spent a few hours on this yestarday. The 1.7.0 download [1] is not autoconf and I was unable to persuade it to cross compile with our toolchain let alone with the nspr. If you have a recipie that would be great... > > Chris. > > > [1] http://ftp.mozilla.org/pub/mozilla.org/js/js-1.7.0.tar.gz -- Regards Vincent http://www.kyllikki.org/