Date: Wed, 30 Aug 1995 11:32:54 -0700 (MST) From: Terry Lambert <terry@lambert.org> To: rgrimes@gndrsh.aac.dev.com (Rodney W. Grimes) Cc: pete@kesa26.kesa.com, freebsd-platforms@FreeBSD.org, pete@RockyMountain.rahul.net Subject: Re: FreeBSD CD-ROM 2.0.5 - Any SPARC Porting Underway? Message-ID: <199508301832.LAA18904@phaeton.artisoft.com> In-Reply-To: <199508300334.UAA05746@gndrsh.aac.dev.com> from "Rodney W. Grimes" at Aug 29, 95 08:34:28 pm
index | next in thread | previous in thread | raw e-mail
> > I wonder if you actually appreciated AT&T makeing a seperate release tape > > for sparc, i386, and other archs. As far as X, I find it nicer to have 1 CD > > set for both SPARC and Intel arch's. I do see the Xfree integration as > > being incomplete. > > Perhaps you should try to find out more about the reasoning that was behind > AT&T's release mechansims that caused them to release the product the way > they did. It was not practical to make a tape that was loadable by multiple > machines (almost impossible for i386 vs anyone else, almost impossible for > sparc vs anyone else, etc). The reason was their source tree layout mechanism. One can only do developement in an extracted source tree, and an extracted source tree has to be architecture dependent. This is why "the keys to the source tree" were so sought after at Novell USG when I was there. It's nearly impossible to do developement that is applicable to more than one platform without involving more than one team, which means interminable meetings. It took me 6 months to increase a reserved area in an unexposed kernel structure by 128 bytes. I kid you not. 8-(. > I am not anti workstation, quite the contrary, I help companies spend billons > of dollars a year in that market, and IMHO, FreeBSD does not belong anyplace > close to that world as we are not going to get the $10K to $100K software > package writers to port to a <$100 OS. Even the likes of BSDI are not > going to ever see any of that market segment, IMHO. With respect, this is why you *MUST* suffort binary compatability with the native OS for the platform when you replace the OS. Supporting a common ABI buys you this, even if the companies don't port to you specificially. One potential market is the one Altos used to have in the x86 Xenix market: Everyone used Altos systems to do developement because they had the most robust environemtn of all the vendors. Altos-generated binaries would run on SCO, Cubix, Interactive, and uPort boxes. SCO binaries would only run on SCO. If you want to be the developement platform of choice (assuming that as a goal), then you ensure that by using your platform, the developers have less work than if using any other platform. Most commercial OS's have their own "unique" interfaces as "value add" which in effect simply unportabalize code. This is to the vendors advantage in that it plays down the commoditization of hardware. Many hardware vendors are only distinguised by the quality of their OS and the applications they've convinced vendors to port -- Sun, HP, DEC. Terry Lambert terry@lambert.org --- Any opinions in this posting are my own and not those of my present or previous employers.home | help
Want to link to this message? Use this
URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?199508301832.LAA18904>