[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: Linux to MacOSX port effort.
From: |
schristley |
Subject: |
Re: Linux to MacOSX port effort. |
Date: |
Sun, 28 Apr 2002 04:27:10 -0700 |
On Thursday, April 25, 2002, at 09:42 AM, Marcus G. Daniels wrote:
"SC" == Scott Christley <address@hidden> writes:
SC> I've been using the stock cc and have been able to compile the
SC> misc directory and part of defobj so far. Is there a specific
SC> reason why the new gcc needs to be used?
The parts of defobj you can't compile are likely the parts that use
nested functions. The Apple GCC 3.1 branch handles a simple case like
this now:
aye, I've disabled hdf5 support at the moment which is where most of
that syntax is used.
I would be concerned that compile-time flags and configuration data
would be wrong if the autoconf tests were not done. It seems like the
easiest way to identify the right configuration details would be to
get the normal build working, and then repackage the details for
ProjectBuilder.
I've run configure and make headers at least, to get the *.xm, classes.h
and types.h files. I did have to modify the Makefile so the
preprocessor didn't munge the files.
SC> The other issue are the two objc runtimes; I'm not sure how
SC> practical using the GNU runtime will be on OSX except for a purely
SC> X approach. I'm still getting familiar with the guts of Swarm
SC> source code so that I can add code to use Apple's runtime.
It isn't possible to use the NeXT-like runtime with Swarm. There are
enhancements added to the (GNU) libobjc included with Swarm that handle
the assembly and disassembly of method call frames.
I believe you are referring to the mframe source code; I recognize it
from gnustep-base, it was the basis for distribute objects. OSX has a
working DO system, so it is just a matter of hooking the method calls to
NSInvocation. From reading the source code, I should be able to
interact with the FCall and FArgument classes, I need to understand
where the entry points are.
By the way, kudos to Marcus and the other developers, the code really is
nicely structured; it has been relatively easy to port. I've a project
now that compiles and links completely, but crashes right after the
modules are initialized. I've left out the gui classes, but I should be
able to do a batch run once i get that far. There is quite a bit of
important code that I have commented out, so it will take some coding
before the program actually works.
The areas that need work include objc class methods, type encoding, and
variable args; all of these exist on OSX but the api's are different
from GNU.
cheers
Scott
==================================
Swarm-Support is for discussion of the technical details of the day
to day usage of Swarm. For list administration needs (esp.
[un]subscribing), please send a message to <address@hidden>
with "help" in the body of the message.
- Re: Linux to MacOSX port effort., W . Northcott, 2002/04/22
- Re: Linux to MacOSX port effort., Perrone Alessandro, 2002/04/23
- Re: Linux to MacOSX port effort., Marc-Antoine Parent, 2002/04/23
- Re: Linux to MacOSX port effort., Marcus G. Daniels, 2002/04/24
- Re: Linux to MacOSX port effort., Scott Christley, 2002/04/25
- Re: Linux to MacOSX port effort., Marcus G. Daniels, 2002/04/25
- Re: Linux to MacOSX port effort.,
schristley <=
- Re: Linux to MacOSX port effort., Marcus G. Daniels, 2002/04/28
- Re: Linux to MacOSX port effort., Scott Christley, 2002/04/29
- Re: Linux to MacOSX port effort., Marcus G. Daniels, 2002/04/30
- Re: Linux to MacOSX port effort., Marcus G. Daniels, 2002/04/30