Date: Sat, 25 Jan 2003 03:40:00 +0100 From: Jeroen Ruigrok/asmodai <asmodai@wxs.nl> To: Terry Lambert <tlambert2@mindspring.com> Cc: JacobRhoden <jrhoden@unimelb.edu.au>, freebsd-chat@freebsd.org Subject: Re: GCC as a selling point for FreeBSD? (Not!) Message-ID: <20030125023959.GO67360@nexus.ninth-circle.org> In-Reply-To: <3E30EF41.D4E26A23@mindspring.com> References: <4.3.2.7.2.20030119130825.00b21ee0@localhost> <4.3.2.7.2.20030119133833.00e422f0@localhost> <200301201620.37863.jrhoden@unimelb.edu.au> <3E2B9C4C.8626D11C@mindspring.com> <20030124071422.GK67360@nexus.ninth-circle.org> <3E30EF41.D4E26A23@mindspring.com>
next in thread | previous in thread | raw e-mail | index | archive | help
-On [20030124 11:25], Terry Lambert (tlambert2@mindspring.com) wrote: >Jeroen Ruigrok/asmodai wrote: >OK... The C++ producer internally compiles to TDF, which is a >variant of [X]ANDF, according to the documentation, which, if >you look at the definitions in the Aho "Compiler Design" book >qualifies as compiling to a quad tree as an intermediate. By your reasoning, salt may "qualify" as sugar, but don't be surprised if people refuse to eat what you cook. >So what is it you were objecting to, exactly? Me calling the >intermediate format by it's generic technical name, instead of >the brand specific name of "TDF"? You abusing terms for non-correct explanations. >My point was only that it's possible to "tunnel" information, >like "asm" statements, should you wish to extend the compiler >to support them, through to the back end converter... it's >not "impossible", which was what was being claimed. Doing inline assembly in something like TenDRA is not impossible, but it requires good thought how you will deal with it. Personally I still find seperate .s files combined with .c/.h files the best. >> Furthermore, to quiet some other speculation and such on this topic: >> TenDRA is BSD Licensed, we're actively working together with some >> committers to get gcc-dependencies fixed in the source tree (on another >> note, we also periodically test apps: >> http://www.tendra.org/~asmodai/compiled-apps.txt). > >We've heard Brett Glass rail on it long enough that we know the >license is acceptable, even if the code generation and optimizer >are not really very adequate. What part of BSDL was not apparent on the website's frontpage or my text above? The DERA [Qinetiq] folks were kind enough to explain their Crown license is a BSDL-like license and had no objections to use putting it under BSDL for the new work. So, if the BSDL/Crown duality, for now, is not acceptable as a license I wonder what would be within the scope of the FreeBSD Project. So I fail to see the relevence of the above comment to Brett. >I think the major issues, as far as "C" is concerned, are the >GCC-ism's, like auto array declarations using an index that was >a parameter to the function, which still occurs one place in the >FreeBSD kernel, and the inability to support "Linker sets", per >se, in order to support SYSINIT() type constructs, without the >ability to support inline assembly code -- and supporting the >inline assembly code, itself. Correct, FreeBSD depends heavily on gcc abuses/extensions of the C language, such as __attribute__(()), #warning, and so on. >>From what I was able to see in the published documentation on >the site, there is still no support for inline assembly. I was >unable to get an answer, one way or another, whether the code >currently supports "long long", or not, either. It has supported long long since 4.1.2 which was released in 1998. -- Jeroen Ruigrok van der Werven <asmodai(at)wxs.nl> / asmodai / Kita no Mono Ninth Circle Enterprises | WTO + WIPO = DMCA? http://www.tendra.org/ | http://www.anti-dmca.org/ Man is the Dream of the dolphin... To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-chat" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20030125023959.GO67360>