On 28/03/11 12:28 PM, Brian Davis wrote: > David G. Koontz wrote: >> >> Only one problem, it doesn't generate code (no object files). >> > It looks like you're building an mcode target, which doesn't create object > files. > I sort of got side tracked looking at fauhdlc. It wouldn't build under 10.6.7 with Macports providing tools due to an internationalization dependency on a lex specification used by flex which Linux cured by default. flex 2.5.35 uses a different set of dynamic libraries under Macports than are found in it on Linux and wouldn't swallow UTF-8. I came up with a work around and a fix and today the author provided the patched lex specification to remove the dependency that's going into the next release. He was also kind enough to look at my test-run results and verify they were all known problems.
I had noticed that procedure didn't build all the ghdl targets. Also the README tells you, you need to build from the source distribution included into the gcc tree first. I've done that exactly once on a Mac and the recipe is obsolete. I've always intended to do a Macports port but no one seems to like the idea of require installation of a Macports environment. I'm not big on Xcode environments myself, ghdl isn't a GUI focused tool. I can extract the configuration needed from Macports to build it stand alone. It may be worth doing it under Xcode, which is the root of any build environment. I'd hate to get embroiled in version issues there, though. >> From your earlier post: >> >> Macbook: ghdl_mcode --version >> GHDL 0.29 (20100109) [Sokcho edition] >> Compiled with GNAT Version: 4.4.0 20080314 (experimental) >> mcode code generator >> > > Does a simple "hello world run" without the segmentation fault? It wouldn't get elaborated without the compiler. It doesn't produce any error message for not finding ghdl1. > > It might be worth trying a build with the Mac version of GNAT GPL: > http://libre.adacore.com/libre/download/ > > Note that the only free Mac GNAT GPL version is an x86_64-darwin target. > ( I tried a build yesterday, but the GNAT 64 bit binaries don't run on my > older mini with a "Core Duo" CPU. ) > They should run under darwin9.6 or darwin10 environments (10.5.7 and up if memory serves). A Core 2 Duo is an x86_64 machine, it sounds like you have a 2006 era machine. Another thing to try would be bootstrapping gnat through gcc-Ada using the MacAda tool (it's universal) and building something with i386 support. (The gcc/ada subdirectory contains source written in Ada). I spent Friday updating my Macports listening to the fan howl (QT3 and QT4). It's a lot of work keeping up to date. Mac OS X moves fast, Xcode almost as fast, Macports even faster. All the references someone provides as a how to tend to get outdated quickly. It seems every way there involves research. Who would have though you'd need to resort to archeology after only a couple of years. It's enough to make you wish for interpretive tool environments without dependencies. Ghdl seems to need platform champions. We've seen the Redhat guys struggle in the past. The problem isn't unique to Mac OS X, and certainly not to ghdl. _______________________________________________ Ghdl-discuss mailing list [email protected] https://mail.gna.org/listinfo/ghdl-discuss
