Re: [Tools-arch] Document Production Methods

[Martin Thomson <[e-mail]>]

On Wed, Sep 2, 2020, at 12:36, John Levine wrote:
> I think we underestimated how hard it is to do automated typsetting
> that produces good looking results so we ended up with a whole lot of
> very complex tagging to let people say in great detail how they want
> things to look. We also put in a lot of semantic tags, e.g., this is a
> picture, that is sourcecode, this other thing is someone's name, which
> may be useful for future automated processing but we don't know yet.

I guess what I'm challenging here is the inherent value of those annotations.

For the typesetting stuff, as that is tied to a very specific instance, attempting to preserve those as text is subsequently changed might be counterproductive.

For the semantic stuff, as they are not the product of any negotiation/consensus process, it is unlikely that they have any special status.  As such, I suggest that they exist on much the same level as other unofficial markings.  Useful or not, they aren't representative of the artifact we are looking to preserve.  They are only useful to the extent that they allow us to better communicate about that artifact, to which you next point might be the most salient in this regard...

> One of the few things that XML makes simple is to tell what is a tag
> and what is text.

That unfortunately isn't always respected, as I recently learned.  But that's a tooling issue that is easily rectified.