Re: Missing keyword 'tied'

[Peter Stephenson <p.w.stephenson@xxxxxxxxxxxx>]

> On 04 January 2022 at 16:22 Jim <linux.tech.guy@xxxxxxxxx> wrote:
> Your patch triggered something.  The keyword 'special' for the parameter
> flag 't' is defined as "for special parameters defined by the shell".
> Should it
> also include user-defined parameters? Case in point, parameters used by
> ztie to "tie" to a gdbm file. Technically, the parameter itself is defined
> by the
> user, yet it has the 'special' keyword.  Sorry if this seems to be
> nitpicking,
> but 57 years of computer logic it's hard to break the habit.
> 
> Also when I first started using ztie I expected the parameter to include the
> key word 'tied'. Had to train myself to check parameter zgdbm_tied.

In this case the generic parameter code is getting confused by the details
of the ztie interface, so that would take some special casing to sort out ---
unless we simply document that ztie creates special parameters, which is
probably the way to go.  If the (t) flag says they're special, they're special,
no matter how they were created; it's the same flag that the shell uses everywhere
to test this.

"Tied" in the sense in the generic parameter code has a different meaning, which
doesn't apply to database ties (we should have been more careful about the language).
I don't think it's possible for the (t) code to detect the existence of a database
link without doing a lot more work than it's designed for.

pws