On 8/9/07, Tim Bray <[EMAIL PROTECTED]> wrote: > On Aug 8, 2007, at 12:11 PM, Bruno Jargot wrote: > > > I thought we're talking about making /bin/sh being ksh93, NOT bash > > Using bash would be the wrong direction. Indiana should make thinks > > BETTER than Linux. > > > > A few numbers: > > This all feels a little surreal. Remember, we're targeting developers. > > 1. the *n*x shell is *not* a competitive differentiator. Nobody > cares if a shell is "better". Nobody. > 2. shell performance is irrelevant.
Shell performance is relevant. It matters if the wrapper script for Firefox or Starofffice needs 0.2 (ksh93) or 4 seconds (bash) to execute. It matters if Indiana can avoid using perl during the boot process because anything can be done in one shell instead. It matters if installing ksh93 as /sbin/sh saves 11 seconds during the boot process. Shell performance is relevant because the shells are the way how Unix applications are connected. > The things that will make Indiana better in a way that developers > will notice will be better/safer package management, better > observability (dtrace), better data management (zfs), and so on. For > totally commoditized things like /bin/sh, the decision procedure > should be: > > 1. Do a little research and find out what the highest proportion of > target developers is used to > 2. Choose that. If you prefer bash - which version are you preferring? bash2, bash3 or the upcoming bash4? bash3 broke many bash2 scripts and bash4 will break some of the existing bash3 scripts. I do prefer to go with ksh93 in this case - it has a test suite which guarantees backwards compatibility for it's scripting language since 1994 and is closer to Solaris backwards compatibility guarantee than the chaos caused by each major release of GNU/bash. Irek _______________________________________________ indiana-discuss mailing list [email protected] http://mail.opensolaris.org/mailman/listinfo/indiana-discuss
