John Plocher wrote: > Glenn Skinner wrote: > > (latin) alphabet for its version number, e.g. version 'a', > > version 'b' etc. ; the '+'/'-' means the stabilty status, e.g. > > '-' means its "alpha", no suffix means its "stable" (e.g. ready > > for production usage) and '+' means its a bugfixed stable version > > etc.). > > > > Are we to infer from this nomenclature description that ksh's > > stability level is decreasing as part of this case? > > I think there may be a definition overload here: > > stability(ARC) => interface evolution over time > -vs- > stability(AST/KSH) => quality and bug levels > > I don't believe the interface taxonomy stability levels > are changing at all, but the implied "quality of the release" > metric is.
Right... we're talking about version numbers, not ARC interface stabilty levels... .. some notes: 1. Orginally we planned to use ksh93 version "s+", however the __MASSIVE__ amount of bugfixes and other changes needed required at some point so many changes that it wasn't suiteable for the ksh93s+ stable branch anymore. However as a result we have all matching RFEs and bugfixes implemented (including the request for "static" variables in functions, a _major_ set of bugfixes for variable tree handling (which required a full rewrite of ksh93's database-like variable storage subsystem) and lots of other tweaks and changes) and that's IMHO more worth than crouching around with a stable version and still having bugs open in bugster. 2. Just because the code is marked as "alpha"/"beta" doesn't mean it's "instable". To ensure a certain level of stabilty of the code we run the ksh93 test suite (currently: 41 tests, however four more are currently being under construction) for each locale installed on the system ($ locale -a | wc -l # says B84 has 332 locales installed, we use a subset of ~~90 locales right now (which means a single test suite cycle alone now runs more than two hours on a decent system)) and the new putback will even tighten the grip on any bugs by using a a) secret switch in libast's memory allocator system to do libumem-style memory corruption tests and b) testing both normal script code and compiled script code. Additionally we run manual tests and run the POSIX test suite. And we release binaries for testing and make sure that using ksh93 as /sbin/sh+/usr/bin/sh results in a bootable system which can compile OS/Net. Or short: We do massive amount of testing - please don't worry about the stabilty of the code (which doesn't mean we're 100% secure against any isues we don't know about (which means: Please test the test binaries and _report_ bugs or any weired issues. We can't fix bugs if we don't know them...)). ---- Bye, Roland -- __ . . __ (o.\ \/ /.o) roland.mainz at nrubsig.org \__\/\/__/ MPEG specialist, C&&JAVA&&Sun&&Unix programmer /O /==\ O\ TEL +49 641 7950090 (;O/ \/ \O;)
