From mboxrd@z Thu Jan 1 00:00:00 1970 To: 9fans@cse.psu.edu Date: Wed, 6 Sep 2000 08:45:04 +0000 From: Christopher Browne Message-ID: References: <39B4C71E.CEA88DDB@ysbl.york.ac.uk>, Subject: [9fans] Re: Kernighan interview (w/ Plan 9 mention) Topicbox-Message-UUID: 037f2e6e-eac9-11e9-9e20-41e7f4b1d025 Centuries ago, Nostradamus foresaw a time when Conway Yee would say: >Leo Caves writes: >> Arguably, Linux transitioned from its hobbyist niche >> to a wider acceptance through a server role. The effort now >> seems to be back to the desktop. >> >> Its difficult to tell in what way Plan 9 might make such a transition. > >I would argue that such a transition will likely never take place. > >First, there is usually only room for 1 "killer app" in the market and >Linux has already taken up that role. _Partially_ true... The "killer app" that I'd consider Linux provides is a "freely licensed, freely modifiable environment in which to run All Sorts Of Unix Code." That's not "so killer" as to have prevented there from being quite viable teams working on FreeBSD, NetBSD, and OpenBSD, that provide much the same "killer app." The problem for Plan 9 is that it _isn't_ quite such a platform. It's _not_ "more-or-less a Unix clone." While there are common roots with Unix, the way Plan 9 proves really useful is if it is used to do things you couldn't do with Unix. There is a close parallel here with Hurd; the major _merit_ of Hurd, over Linux, is the ability to do Cool Stuff using the Hurd translators that provide cool mappings of system facilities onto filesystems. The problem with this is that if you write applications that depend on "Cool Stuff" like Hurd translators or Plan 9 namespaces, those applications can't run anywhere else. Which makes them "niche" apps, which will be largely ignored by the large proportion of the population that use more "mainstream" systems. You may have a cool HTTP server where all the behaviour/configuration is exposed as a virtual filesystem; if it only runs on Plan 9, interest will be limited to the point of extinction. >Second, who can forget the litigation over NET/2? Anyone who >contributes with the intent of transitioning to a server role will >eventually have to deal with ATT's lawyers. I believe that the >engineers/scientists at ATT are honorable but are their lawyers? > >Third, as it stands, Plan 9's license is hardly appropriate to those >who would run it for serious applications. Who wants to take the risk >of having ATT own your application? This just makes things worse; even if the lawyers _are_ honorable, nobody knows for sure until _after_ they deploy the applications, and the uncertainty will be crippling. -- (concatenate 'string "aa454" "@" "freenet.carleton.ca") Rules of the Evil Overlord #83. "If I'm eating dinner with the hero, put poison in his goblet, then have to leave the table for any reason, I will order new drinks for both of us instead of trying to decide whether or not to switch with him."