.TH EM_B 6 2017-01-18
.ad
.SH NAME
em_b \- ACK B compiler
.SH SYNOPSIS
.B ~em/lib/ack/em_b
.RI [ options ]
.SH DESCRIPTION
.I em_b
is a port of the ABC B compiler to the ACK.
Interested parties will be
interested in the upstream distribution here:
.nf
.sp
https://github.com/aap/abc
.fi
.PP
However, the version here has been heavily modified \(em bug reports should be
filed with the ACK, not with the upstream compiler.
.PP
Since B was designed for machines with word addressing, some hacking is
required to make it work on modern, byte addressed machines.
The generated
code expects B variables to contain word addresses, and then generates
code to transform these into native addresses before use (which,
unfortunately, impacts performance).
However, the ACK's linker doesn't know
how to emit word addresses into the program's data sections, and so a
separate fixup stage has to happen at runtime, just before \fBmain()\fP,
to convert the byte addresses into word addresses.
.PP
The end result is that using multiple source files with B is somewhat
unwieldy, requiring each module to be explicitly named and then an extra
stage to generate the fixup code.
See the \fBack\fP(1) and \fBabmodules\fP(1) for details.
.SH OPTIONS
.I em_b
accepts the following flags:
.IP \-w\ \fIsize\fP
Sets the word size, used for scaling addresses.
Usually either 2 or 4.
.IP \-B\ \fIname\fP
Sets the name of the module currently being compiled (used to generate the
fixup table symbol name).
Defaults to \fImain\fP if not specified.
.IP \-i\ \fIfilename\fP
The source B file.
.IP \-o\ \fIfilename\fP
The output compact EM bytecode file.
.SH SEE ALSO
\fIack\fR(1), \fIabmodules\fR(1)
.SH REMARKS
It is very unlikely the \fIem_b\fP will ever be useful for anything.