On Sat, Aug 30, 2025 at 03:51:49AM +1000, Martin D Kealey wrote: > On Sat, 30 Aug 2025, 02:26 Chet Ramey, <[email protected]> wrote: > > > There is no specification for how it behaves; > > > I'll buy that when the shell is in POSIX mode, but when talking about > Bash's extensions to POSIX, there's never been a full specification > separate from the source code > > there is no expectation that > > it will remain unchanged in the future > > > That's your expectation, but it's not many users' expectation, when they > have decades of experience to the contrary. > > You might say that the source code is always the official reference; I > > don't buy it here. > > > Not the source code per se, but its behaviour. > > Most of what I initially learned about Bash was by reading the manual, then > systematically experimenting to establish corner cases. There wasn't an > alternative if I wanted to write reliable scripts back then; there were far > too many holes in the documentation. > > "It was never documented" is a pretty weak argument when significant parts > of the shell's seemingly intended behaviour are STILL undocumented. So > users will *of course* regard the detectable behaviours of the source code > as the authoritative language definition, whether or not they describe it > in those terms. > > -Martin
When I worked for Prime Computer, we used the terms "documented" and "supported" interchangeably. If a feature was documented, it was supported. If a feature was not documented, it was not supported. That has been my expectation ever since. In fact, an undocemented feature was removed, that had been used by an important customer. We made no apology for that. I'm with Chet on this. Cheers ... Duncan.
