Fixes necessaries for compile dfly with clang

Juan Francisco Cantero Hurtado iam at juanfra.info
Thu Nov 17 13:32:27 PST 2011


On 11/17/2011 08:11 AM, John Marino wrote:
Other problem is the C++ code (e.g. binutils doesn't compile) but
I consider most important now only the problems related to C.


Well, that's a deal-breaker. If the project made a decree that the
base compiler languages were limited to c-languages, then I would
probably advocate clang over gcc in the long-term (I know Samuel's
argument, and I could provide some rebuttal to his concerns).
However, that means we accept C++ in the base, and I would like to
see the gold linker be able to be a drop-in replacement for ld (I
have hopes for binutils 2.22).  If clang can't property build a
fundamental part of the toolchain, it's a non-starter in my eyes.
That said, I have a hard time believing the clang folks would let
binutils not be buildable.  What's wrong there?
Don't worry. I talked about the C code because the most of the code in
dfly is written in C, so this should be the first step. The fact is I
would like a base only with C code or with more languages available like
python, perl and gccgo :) (I think I'm dreaming). I think the arguments
from you and Samuel are correct both.
Compile dfly with two different compilers is very good for the project,
e.g.: clang cried like a baby because the bug in the Makefile of
libmagic but gcc ignored the problem.
I don't remember what was the problem with binutils. It's strange
because clang also can use the gold linker. But citrus is more important
now, so step by step.
Cheers.

--
Juan Francisco Cantero Hurtado http://juanfra.info




More information about the Kernel mailing list