On Fri, Sep 20, 2002 at 04:04:04PM +0200, Lars Gullik Bj�nnes wrote:
 
> Andre Poenitz <[EMAIL PROTECTED]> writes:
> 
> | On Fri, Sep 20, 2002 at 10:24:52AM -0300, Joao Luis Meloni Assirati wrote:
> >> Right, and this button already exists. [...]
> >
> | We have four cases:
> >
> |  (1)   validate  finds AMS,              user wants it
> |  (2)   validate  finds AMS,              but user does not want it
> >
> |  (3)   validate  does not find AMS,      but user wants it
> |  (4)   validate  does not find AMS,      user does not want it
> >
> >> The idea is that that button controls the use of AMS, and two "persons"
> >> can check it: the user and validate, but only the user can uncheck it.
> >
> | (4) is not covered by your proposal.
> 
> But isn't (4) then a degenerate case?

No, here Andr� is right. The problem is that validate isn't run *until
the latexing phase*, at which time it is too late for user intervention.

It should be possible for the user to specify *in advance* what should
happen

(a) if validate finds AMS; or
(b) if validate does not find AMS.

The three-state radio button 
(use AMS always/ use AMS never/ trust the validation) offers this. 
(And of these, "use AMS always" can be replaced by a preamble entry
as it will be for experts only.)

> -- 
>       Lgb

Martin

Attachment: msg45089/pgp00000.pgp
Description: PGP signature

Reply via email to