On Wed, 2005-07-27 at 23:31, Helmar Wodtke wrote: > I'm really impressed by the discussion about ksh. But I dont get the point. the point is that scripts (not users) might rely on features (and bugs) as exhibited by ksh88. if you replace it with some other shell (different bugs, different features), those scripts might not work anymore as intended.
and that's bad, if you made a promise on the stability of that interface. of course, opensolaris could claim it never had any promise on ksh88 availability, so it doesn't have to provide it (and that interface would be a sun solaris-only property). then sun could decide to keep this interface, or to deprecate and remove it, to reconcile that difference. or, opensolaris gets some ksh88 compatible shell (by forking pdksh, or by whichever other means), and after that's done, ksh88 becomes part of the opensolaris interface - hopefully including a promise about its stability. making promises on APIs, and keeping them, is unfortunately a pretty weak spot of many open source projects so far, and worse, some even play the "embrace/extend/extinguish" game they accuse others of. (extending beyond the minimum subset can be good, but there should always be a way to work with the standard set only, POSIXLY_CORRECT was a nice idea, just lacking a working implementation) discussing about interfaces and standards is a good thing, standards are a good thing, the hack&slay mentality often shown in the so-called "open source community" is not. patrick mauritz _______________________________________________ opensolaris-discuss mailing list opensolaris-discuss@opensolaris.org
