This is the mail archive of the gcc@gcc.gnu.org mailing list for the GCC project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

alternative for bootstrapping non-C frontends


Hi folks,

after reading mails on this list about people having to jump through 
hoops to bootstrap the ADA front-end on non-mainstream architectures, 
I had an idea for bootstrapping GCC that I would like to present and ask 
your feedback upon...

Would it be possible to distribute RTL releases of the non-C frontends 
and compile those as part of stage1 bootstrapping?

I understand that it is impossible to distribute binaries for every 
possible architecture + os + runtimelibs... distributing an RTL version 
of the compilers would not have this problem, as it is platform-independent.

Let me clarify:

If the frontends could be compiled to RTL, and some external 
representation of that platform-independent representation of the 
programs be written to files (eg. .rtl instead of .o files), then one 
could get away with bootstrapping the C frontend, and the backend for 
the platform, and then compile the .rtl files to .o files, and link 
those to obtain a working frontend. This would remove the need to find 
an existing ADA compiler for the target platform, or the need to go 
through the process of building a cross-compiler.

This process would simplify bootstrapping for compilers that are not 
written in C.

I don't know if this has been proposed before, and if it has, I hope 
this is not bait for a holy war on the list...

-- lode


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]