Re: Allow unknown steps?

/ Jeni Tennison <jeni@jenitennison.com> was heard to say:
| Norman Walsh wrote:
|> I wonder if we should say instead that it's an error to attempt to
|> evaluate a step that the implementation does not know how to perform.
|> Then the above pipeline would not be an error. It would also not be an
|> error to use p:system-property to make choices based on XProc vendor
|> information.
|
| I'd rather use a p:step-available() than p:system-property() (since
| what individual implementations can do usually changes more quickly
| than I imagine updating my pipelines).

Yes, if we make this change, I think p:step-available() is a better
answer than system properties.

Note, however, that I would consider "use-when" to be feaping
creaturism and would object violently.

| Another alternative is to use <p:try>/<p:catch>, assuming that
| <p:try>/<p:catch> can catch dynamic errors?

It can. You could. But I think that'd be pretty tedious. Or not, hard
to say. Anyway, assuming we go this way, I don't think step-available
is very expensive.

                                        Be seeing you,
                                          norm

-- 
Norman Walsh <ndw@nwalsh.com> | No one gossips about other people's
http://nwalsh.com/            | secret virtues.--Bertrand Russell