forked from pallet/ritz
/
HACKING
158 lines (114 loc) · 5.99 KB
/
HACKING
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
* The SLIME Hacker's Handbook -*- outline -*-
* Lisp code file structure
The Lisp code is organised into these files:
swank-backend.lisp:
Definition of the interface to non-portable features.
Stand-alone.
swank-<cmucl|...>.lisp:
Backend implementation for a specific Common Lisp system.
Uses swank-backend.lisp.
swank.lisp:
The top-level server program, built from the other components.
Uses swank-backend.lisp as an interface to the actual backends.
slime.el:
The Superior Lisp Inferior Mode for Emacs, i.e. the Emacs frontend
that the user actually interacts with and that connects to the
SWANK server to send expressions to, and retrieve information from
the running Common Lisp system.
contrib/*.lisp:
Lisp related code for add-ons to SLIME that are maintained by
their respective authors. Consult contrib/README for more
information.
* ChangeLog
For each change we make an entry in the ChangeLog file. This is
typically done using the command `add-change-log-entry-other-window'
(C-x 4 a). The message can be automatically extracted from the
ChangeLog to use in a CVS commit message by pressing C-c C-a in a
vc-mode or pcl-cvs commit buffer.
ChangeLog diffs are automatically sent to the slime-devel mailing list
each day as a sort of digest summary of the slime-cvs list.
There are good tips on writing ChangeLog entries in the GNU Coding Standards:
http://www.gnu.org/prep/standards/html_node/Style-of-Change-Logs.html#Style-of-Change-Logs
For information about Emacs's ChangeLog support see the `Change Log'
and `Change Logs and VC' nodes of the Emacs manual:
http://www.gnu.org/software/emacs/manual/html_node/emacs/Change-Log.html#Change-Log
http://www.gnu.org/software/emacs/manual/html_node/emacs/Change-Logs-and-VC.html#Change-Logs-and-VC
* Sending Patches
If you would like to send us improvements you can create a patch with
C-x v = in the buffer or manually with 'cvs diff -u'. It's helpful if
you also include a ChangeLog entry describing your change.
* Test Suite
The Elisp code includes a command `slime-run-tests' to run a test
suite. This can give a pretty good sanity-check for your changes.
Some backends do not pass the full test suite because of missing
features. In these cases the test suite is still useful to ensure that
changes don't introduce new errors. CMUCL historically passes the full
test suite so it makes a good sanity check for fundamental changes
(e.g. to the protocol).
Running the test suite, adding new cases, and increasing the number of
cases that backends support are all very good for karma.
* Source code layout
We use a special source file layout to take advantage of some fancy
Emacs features: outline-mode and "narrowing".
** Outline structure
Our source files have a hierarchical structure using comments like
these:
;;;; Heading
;;;;; Subheading
... etc
We do this as a nice way to structure the program. We try to keep each
(sub)section small enough to fit in your head: typically around 50-200
lines of code each. Each section usually begins with a brief
introduction, followed by its highest-level functions, followed by
their subroutines. This is a pleasing shape for a source file to have.
Of course the comments mean something to Emacs too. One handy usage is
to bring up a hyperlinked "table of contents" for the source file
using this command:
(defun show-outline-structure ()
"Show the outline-mode structure of the current buffer."
(interactive)
(occur (concat "^" outline-regexp)))
Another is to use `outline-minor-mode' to fold away certain parts of
the buffer. See the `Outline Mode' section of the Emacs manual for
details about that.
(This file is also formatted for outline mode. If you're reading in
Emacs you can play around e.g. by pressing `C-c C-d' right now.)
** Pagebreak characters (^L)
We partition source files into chunks using pagebreak characters. Each
chunk is a substantial piece of code that can be considered in
isolation, that could perhaps be a separate source file if we were
fanatical about small source files (rather than big ones!)
The page breaks usually go in the same place as top-level outline-mode
headings, but they don't have to. They're flexible.
In the old days, when slime.el was less than 100 pages long, these
page breaks were helpful when printing it out to read. Now they're
useful for something else: narrowing.
You can use `C-x n p' (narrow-to-page) to "zoom in" on a
pagebreak-delimited section of the file as if it were a separate
buffer in itself. You can then use `C-x n w' (widen) to "zoom out" and
see the whole file again. This is tremendously helpful for focusing
your attention on one part of the program as if it were its own file.
(This file contains some page break characters. If you're reading in
Emacs you can press `C-x n p' to narrow to this page, and then later
`C-x n w' to make the whole buffer visible again.)
* Coding style
We like the fact that each function in SLIME will fit on a single
screen (80x20), and would like to preserve this property! Beyond that
we're not dogmatic :-)
In early discussions we all made happy noises about the advice in
Norvig and Pitman's _Tutorial on Good Lisp Programming Style_:
http://www.norvig.com/luv-slides.ps
For Emacs Lisp, we try to follow the _Tips and Conventions_ in
Appendix D of the GNU Emacs Lisp Reference Manual (see Info file
`elisp', node `Tips').
We use Emacs conventions for docstrings: the first line should be a
complete sentence to make the output of `apropos' look good. We also
use imperative verbs.
The biggest problem with SLIME's code base is feature creep. Keep in
mind that the Right Thing isn't always the Smart Thing. If you can't
find an elegant solution to a problem then you're probably solving the
wrong problem. It's often a good idea to simplify the problem and to
ignore rarely needed cases.
Remember that to rewrite a program better is the sincerest form of
code appreciation. When you can see a way to rewrite a part of SLIME
better, please do so!