graphene/LibOS/shim/test/apps/lmbench/lmbench-2.5/doc/graph.1

.\" $Id$
.de DS
.	sp .5
.	nf
.	in +4
.	ft CW
.	vs -1
..
.de DE
.	sp .5
.	fi
.	in
.	ft
.	vs
..
.TH GRAPH 1
.SH NAME
graph \- compile graphs into pic input
.SH SYNOPSIS
.B graph
[ options ]
[
.I filename
\&.\|.\|.
]
.SH DESCRIPTION
.LP
.B graph
is a perl script which
takes sets of X Y data and generates a (human readable) pic program
that will produce the graphed data.  The output is designed such that
you can save it in a file and tweak it to make it fit your document.
Try one and look at the output.  The output is actually commented.
.LP
The graph is autosized and auto ticked.
.LP
The input data format is similar
that of xgraph(1), i.e.,
.DS
1 1
2 2
3 3
"sloped across

1 4
2 4
3 4
"straight across
.DE
.SH "CONTROL OPTIONS"
.LP
You may set the graph title, the X title, and the Y title with the 
following control sequences in the data stream:
.DS
%T Graph title in +4 point font
%X X axis title and/or units in +2 point font
%Y Y axis title and/or units in +2 point font
%fakemax-X <value>     force graph to be that big
%fakemax-Y <value>     force graph to be that big
%fakemin-X <value>     force graph to be that small
%fakemin-Y <value>     force graph to be that small
.DE
.SH OPTIONS
.IP -rev 12
reverse X/Y data sense (and titles).  Note this is done after processing
any fudging of the input data stream(s) (see -xk, -yk, -logx, etc below).
.IP -below
put data set titles below the graph rather than to the right.
.IP -close
no extra space around the data's endpoints.
.IP -qline
connect the quartile center points.
.IP -grid
dotted line grid marks.
.IP -nobox
no box around whole graph.
.IP -big
make the graph take the whole page.
.IP -medium
make the graph take about 1/2 the page.
.IP -small
make the graph be small.
.IP -grapheach
draw each data set in its own graph.
.IP -nolabels
no X/Y/Title labels.
.IP -nodatal
no data set labels.
.IP -nomarks
do not mark each data point with distinct markers (endpoints are still
marked).
.IP -k
print values larger than 1000 as value/1000.
.IP -xk
multiply X input by 1024 (blech).
.IP -yk
multiply Y input by 1024 (blech).
.IP -xm
multiply X input by 1024*1024 (blech).
.IP -ym
multiply Y input by 1024*1024 (blech).
.IP -logx
convert X input into lag base 2 of X input.
.IP -logy
convert Y input into lag base 2 of Y input.
.SH EXAMPLE
Workstation price performance from a Digital ad.  Process with
.DS
.ps -2
graph -rev workstations | groff -TX75

"%T Workstation Price / Performance, 6/93
"%X SPECINT 92 Performance
"%Y Price in $1000's
35 5
65 10
78 15
110 70
"Dec AXP line

25 4
25 8
38 16
48 21
52 23
64 27
"Sun SPARC line
.DE
.ps
.SH "QUARTILE FORMAT"
Data points are \f(CBx y1 y2 y3 y4 y5\fP.   You get a two lines from the
first two y values, a mark at the third, and another line from the last two.
.SH "SEE ALSO"
.BR gtroff (1),
.BR gpic (1),
.BR perl (1).
.SH BUGS
This should probably be called pic_graph or something like that.
.LP
This isn't done as much as I would like.
It isn't integrated with the groff preprocessor yet.
It doesn't know about .GS/.GE things.  I use it to manually generate
a pic file and then include that.