>
> In message <9502162115.AA05327@sqarc.sq.com>, yuri@sq.com writes:
> >
> >when you actually so the substitution for HTML-0, you
> >end up with:
> >
> ><!ENTITY % head.content "TITLE & ISINDEX? & BASE? & " >
> >
> >that is: You have a dangling ampersand.
> >
> >I can't believe I'm the first person to run into this,
>
> You are. My Q/A procedures have become somewhat lax. When
> I fixed the entity stuff at Mr. Grosso's suggestion, I forgot
> to test the level 0 DTD.
>
[..]
>
> I don't think this is legal. I don't believe you can start a parameter
> entity with &, as Mr. Grosso pointed out.. That's what prompted me to
> make these changes in the first place.
>
> Anybody got a legal way to fix this problem?
>
I forgot about this. It would be worth proposing a fix for the SGML revision.
I suspect the cleanest way to do this is to actually repeat the full
declarations within each of the relevant marked sections. Unfortunately
to essentially act as the base level (that is, the way that html.dtd
comes today).
I'll try this now.
If anyone has a better way to solve it, don't hesitate ...
Yuri