Re: Blogging sucks

[prev] [thread] [next] [lurker] [Date index for 2005/10/13]

From: peter (Peter da Silva)
Subject: Re: Blogging sucks
Date: 23:19 on 13 Oct 2005
> * Peter da Silva <peter@xxxxxxx.xxx> [2005-10-13 22:15]:
> > 	What if I'm indenting something that isn't a code block?

> Who cares? HTML has no generic “indent a block” mechanism, so
> that the 

<div style="margin: 0.5i">
I don't quite understand what point you're trying to make here. Not only are
there a number of mechanisms that could be used.
<blockquote>
Traditionally, of course, blockquote has been an option.
</blockquote>
but applying a style to a div is more versatile.
</div>

> Since when are heredocs non-hateful?

Assuming for the sake of argument that a non-hateful solution
exists (which I am not at all sure of), if it's going to allow
you to include verbatim text at all it must be isomorphic to "here
documents", since no other solution actually lets you include
verbatim text.

> You have the choice between
> a terminator that does not appear within the verbatim text (the
> epitome of elegance) or escaping any appearances of the
> terminator within the verbatim text (the paragon of beauty).

You have the choice between a terminator that doesn't appear within
the verbatim text and a terminator that doesn't appear in the
verbatim text. Or you could use a terminator that doesn't appear
in the verbatim text. If you're putting escapes in the text, it's
not verbatim, is it?

> At the beginning of a line within a heredoc the parser is in an
> ambiguous state: what follows may either be another verbatim line
> or may be a hedoc end marker.

In the case of traditional "here documents", the newline is part of the
end marker. In the case of the slashcode "here documents" (which are
hateful for other reasons) the leading and trailing angle brackets are
part of the end marker. In the case of the paired terminators used by IOS
or the "ed/ex" family of text editors the end marker is identical to the
start marker.

> If it’s the former but looks like
> the latter, the parse will disambiguate wrongly, so escaping is
> required. The solution is to require each line to be prefixed
> invidually, so the parser can never guess wrongly by design, so
> that you never have to escape anything.

But then you're not including the text verbatim, you're escaping the
newline. You have to separately modify every line of the text whether
you put the escape at the beginning (as in Fortran) or at the end (as
in the C preprocessor). Which is hateful.

There's stuff above here

Generated at 20:00 on 17 Oct 2005 by mariachi 0.52