Available on GitHub]This was part of a presentation at the Lisp meetup in Seattle circa early 2007 as demonstration of using only standard ANSI Common Lisp functions (thus no need for CFFI, UFFI or implementation-dependent libraries).
A tutorial version of this has been contributed to WikiBooks. While its location has changed over the past ten years, it currently resides under Advanced Topics.
This code repo is merely to make the original files available.
Because this predates Quicklisp, it uses ASDF directly and was only tested with SBCL v1.0.2 on MacOSX 10.4 (Tiger) for PowerPC.
Within the scope of systems programming, there are often multiple languages being used. While there are verbose file formats for exchanging data among foreign applications, sometimes you still must deal with packed structures within “binary” files or network protocols.
This is about processing and generating packed data records such as those native to C programs.
Specifically, we’ll have Lisp reading and writing a C or C++ struct.
We’ll also process Unix time values within ANSI Common Lisp, accounting for the different epochs and depending upon hardware architecture, byte order.
Compare Lisp versus C manipulating 5 bits from a sequence with 7 or more bytes:
Extract
ANSI Common Lisp:
(setf bits (ldb (byte 5 7) value))
C:
bits = (word >> 7) & 0x1F;
Modify
ANSI Common Lisp:
(setf new-value (dpb bits (byte 5 7) value))
C:
new_value = (value & (~(0x1F << 7)))|((bits & 0x1F) << 7);
A better visual is in intro.html or Github CDN version
Some Lisp code here uses character for easily displaying message buffers
from network sockets, which accidentally worked for the presentation but
is wrong.
The proper approach would be to instead use (unsigned-byte 8) as the type
and convert when printing via sb-ext:octets-to-string. For portability,
consider utf8-octets-to-string function in com.google.base package
available via Quicklisp.
As was pointed-out during the presentation: Common Lisp characters contain font information, consume a fairly large footprint in memory, and they are also unlikely to hold necessary range of values that you may need. This limit on values might be unexpected to someone coming from a C programming background. The ANSI CL spec of 1994 requires little more than alphanumerics and control characters be supported, yet contemporary (long before 2017) implementations such as LispWorks and SBCL accommodate UTF-8.
LispSea was restarted in 2007 as a way of giving back to the community for those learning ANSI Common Lisp.
This was in the era shortly after publication of Practical Common Lisp, whereby ANSI Common Lisp was having a bit of a renaissance.
“It’s the environment that I wanted while learning but didn’t exist at the time.”
An earlier version of the group had previously merged with others to become SeaFunc: functional programming. At that time, they typically met for informal discussions over beers.
With blessings from organizers of SeaFunc, from those on LispSea email list, and from those who previously ran LispSea– new meetings were planned.
These meetings were modelled after LispVan– Vancouver, BC. A projector was used during a short presentation that served to seed open discussion and held in a relatively quiet meeting place. In our case, it was held at a community room within a recently updated and expanded library branch.
https://web.archive.org/web/20080517034158/wiki.alu.org/lispsea