Re: Fwd: [PATCH] different approach to --split html

[Eli Zaretskii]

> From: address@hidden
> Date: 09 Nov 2000 19:44:35 +0100
> > 
> > Single-pass operation contradicts several important features (and is a
> > reason for some extremely hairy code), and I wouldn't recommend giving
> > up on those features just to save another pass.  Today's machines are
> > quite fast.
> 
> Well, I guess we disagree on this.  Afaics, texinfo was designed to be
> one pass, and makeinfo too.

Yes and no.  The original makeinfo code did a second pass when it
needed to split the Info output into several files.  Other than that,
it was indeed designed to be one pass.

> I don't know what these ``features'' are

One of them is the ability to have @contents in the beginning, where
God intended it to be.

> but given that they were add-ons, and thus non-essential

I don't agree that anything which was added after the initial release
1.0 is non-essential.  One of these add-ons is HTML output, for
example...

> they require an extra pass, and give lead to ugly code

Actually, the ``ugly'' code was already there, for the most part,
because makeinfo was designed in a way that processing the file and
producing output are closely intertwined, almost like Siamese twins.
There are features, like @macro expansion, which actually produce
expanded text in the output buffer, then pretend the output never
happened by wripping that expanded text out of the output buffer and
resetting the output pointer!

The second pass, when it was added, didn't require any ugly code,
IIRC.  It just reused the existing second pass for splitting the Info
file.