Re: [6562] improvements in index display

[Gavin Smith]

On 23 August 2015 at 22:18, Karl Berry <address@hidden> wrote:
>        \vskip 0pt plus 3\baselineskip
>     -  \penalty 0
>     +  \penalty -300
>        \vskip 0pt plus -3\baselineskip
>
> This plus3 and then plus-3 doesn't make sense to me.
> Did Stepan do that?

Yes he did. I'm not sure I understand it. I think it means to add
positive glue, then allow a break, then if the column is broken, the
positive glue is left in, otherwise it is taken out. Because he
changed the penalty from -300 to 0 at the same time breaking before
index initials was less frequent, so I'm not sure if there were many
cases where this glue did anything.

>
> In any event, as long as you're messing around with index formatting, I
> had an idea that the index might look better if the initials were
> typeset in a box that is two line spaces high.  Then there is a chance
> that the entries might line up side to side.  I'm sure it won't be
> perfect, but maybe it will be better.  Something like:
>
>   \nointerlineskip
>   \vbox to 2\baselineskip{%
>     \vfil
>     \secfonts \secbf
>     \leftline{#1}%
>     \vskip.33\baselineskip % what we have now, maybe should be less
>   }%
>   \nobreak % at following interline space
>
> Of course, any flexible glue anywhere will defeat perfect alignment, and
> I don't suggest getting rid of it, but ... just an idea.

Yes, maybe. Alignment appears to be especially a problem when there is
an initial section at the end, on a page of its own. The columns
frequently don't seem to match up right. I've got an improvement (not
committed yet) where it's possible for the final page of the double
columns only has material in the left column if it's short enough.

One idea I had was to insert a penalty before a final entry under an
initial to stop it being "orphaned"; it doesn't appear to lend itself
to an idiomatic implementation in TeX, though.