Re: info mode header line missing title

[Juri Linkov]

Andreas Schwab <schwab@suse.de> writes:
> "Eli Zaretskii" <eliz@gnu.org> writes:
>> Please show some statistics to back up this assertion.  In my
>> experience, a large percent of Info nodes have long names so that the
>> header line overflows the 79-character limit.
>
> I can confirm this.  On my system more than 50% of the Info nodes would
> overflow the limit if the node name would be included in the header line.
> Without it, only about 10% remain.
>
> $ zgrep -h '^File: .*Node:' *.gz | wc -l
> 16965
> $ zgrep -h '^File: .*Node:' *.gz | sed -n 's/^[^,]*, *//; /.\{80,\}/p' | wc -l
> 9133

This query is not quite right.  The proposed format was without the
"Node: " prefix.  So more proper query is:

  $ zgrep -h '^File: .*Node:' *.gz | sed -n 's/^[^,]*, *Node: //; /.\{80,\}/p' 
| wc -l
                                                        ======

It gives on my system 32% of nodes whose headers would be longer than
80 characters, against 8% of header lines without the node name at all.

But fourfold increase is no good.  So I propose another improvement:
to replace prefixes "Next: ", "Prev: " and "Up: " with 1-character
arrow glyphs representing their meanings ("<" - prev,  ">" - next,
"^" - up).  With this change the header line:

Save Keyboard Macro,  Next: Edit Keyboard Macro,  Prev: Keyboard Macro Query,  
Up: Keyboard Macros

would be displayed as:

Save Keyboard Macro >Edit Keyboard Macro <Keyboard Macro Query ^Keyboard Macros

On my estimations this gives the same percent of nodes (on my system 9%)
as for currently truncated header lines without the node name.
The query to prove it is:

$ zgrep -h '^File: .*Node:' *.gz | sed -n 's/^[^,]*, *Node: //; /.\{99,\}/p' | 
wc -l

since the length of the saved text ", Next:, Prev:, Up:" is 19 characters.

-- 
Juri Linkov
http://www.jurta.org/emacs/