From: John P. <joh...@ho...> - 2002-06-07 13:40:48
|
DEVELOPERS NEEDED IMMEDIATELY TO IMPLEMENT THE FOLLOWING IN THE MINGW32 ALTERNATE C RUNTIME LIBRARY PROJECT: - ANSI/ISO C functions, including all ISO C99 functions and Unicode Functions (integrating parts of it from Newlib, FreeBSD, OpenBSD, and writing other parts from scratch) - Support only native Win32 pathname handling, including UNC pathname handling, except that forward slashes ('/') may be used instead of backslashes ('\') - Patches to GCC to support SEH and Microsoft Platform SDK header files (if necessary, write a patching utility to patch Microsoft Platform SDK header files, especially in the case of the inline assembler) - SEH support in the Mingw32 Alternate C Runtime Library - Patches to Binutils to support Microsoft Platform SDK libraries (and also whole program optimization if possible) Developers must know how to program in C or Intel IA-32 (x86) assembly. - Patches to GCC to support #import directive - Patches to GCC to support implementation of COM objects in C++ without compromising any features required by the ANSI C++ standard - Intel IA-32 (x86) Assembly optimizations to C-Runtime Library functions - Some implementation details can be found at http://mingwacr.sourceforge.net DEVELOPERS MUST KNOW HOW TO PROGRAM IN EITHER C OR IA-32 (x86) ASSEMBLY. DEVELOPERS SHOULD ALSO KNOW HOW TO USE CVS WITH SSH. DEVELOPERS CAN FIND DOCUMENTATION ON HOW TO USE CVS WITH SSH AT THE FOLLOWING URL: http://sourceforge.net/docman/?group_id=1 LIBRARY WILL BE LICENSED UNDER BSD LICENSE. IF YOU WANT TO HOLD THIS POSITION, SEND EMAIL TO THE FOLLOWING ADDRESS: jp...@us... _________________________________________________________________ MSN Photos is the easiest way to share and print your photos: http://photos.msn.com/support/worldwide.aspx |
From: Casper H. <ch...@us...> - 2002-06-07 14:25:39
|
fre, 2002-06-07 kl. 15:40 skrev John Platts: > DEVELOPERS NEEDED IMMEDIATELY TO IMPLEMENT THE FOLLOWING IN THE MINGW32 > ALTERNATE C RUNTIME LIBRARY PROJECT: Your caps lock is stuck. Please write normally. All capitals texts decreases reading speed and is generally annoying to read. > - Patches to Binutils to support Microsoft Platform SDK libraries (and also > whole program optimization if possible) IIRC ld already supports .lib libraries. > - Patches to GCC to support implementation of COM objects in C++ without > compromising any features required by the ANSI C++ standard Please elaborate on that one. > - Intel IA-32 (x86) Assembly optimizations to C-Runtime Library functions Obviously, since I am developing ReactOS, I would like to see the runtime be as portable as possible. What is the reason for using assembly in the runtime? Many of these features would be useful to mingw too so what are the reasons for starting a new similar project? - Casper |
From: Neil B. <ne...@da...> - 2002-06-07 21:15:15
|
John Platts wrote:- > - Patches to GCC to support SEH and Microsoft Platform SDK header files (if > necessary, write a patching utility to patch Microsoft Platform SDK header > files, especially in the case of the inline assembler) I'm not sure I would find the changes to the preprocessor palatable to make this possible. For example, MS's ?$%@&^ header files require the ability to macro-paste two '/'s and comment out the rest of the line. e.g. #define STUPID /##/ STUPID some gunk comments out "some gunk". This is not easily implemented in the current preprocessor, owing to the extreme brokenness of the construct, mixing up the phases of translation (comment removal is done before macro expansion in any sane standards-conforming implementation). I don't particularly want GCC's sources to be kludged to handle such cases. There are other broken things in MS's preprocessor that have been reported to this list in the past, too. > - Patches to GCC to support #import directive GCC already has a #import directive which means something quite different, based on my reading of MS's documentation. The existing meaning of #import already has many users, particularly in the Objective C community. Therefore, personally speaking, I would prefer that the necessary CPP changes were distributed as a patch to GCC's mainline tree, rather than obfuscating GCC just to handle Microsoft's broken header files. Neil. |
From: Zack W. <za...@co...> - 2002-06-07 21:50:11
|
On Fri, Jun 07, 2002 at 10:14:43PM +0100, Neil Booth wrote: > I'm not sure I would find the changes to the preprocessor palatable to > make this possible. For example, MS's ?$%@&^ header files require the > ability to macro-paste two '/'s and comment out the rest of the line. I agree that this, and other outright bugs in their compiler, are not something we wish to support. It would be better to deal with them in the fixincludes framework. > > - Patches to GCC to support #import directive > > GCC already has a #import directive which means something quite > different, based on my reading of MS's documentation. The existing > meaning of #import already has many users, particularly in the > Objective C community. It appears to be totally unambiguous from the code when the Microsoft semantics are desired, instead of the existing semantics; however, the amount of code required to implement Microsoft's semantics would be substantial, and inappropriate for inclusion in the compiler itself. I would be willing to consider a patch that caused us to recognize an attempt to #import a DLL, possibly with additional directives after the pathname, and invoked a separate executable to process the request. That program would produce an appropriate sequence of declarations which GCC could read in as text, via a pipe or temporary file. zw |