aboutsummaryrefslogtreecommitdiffstatshomepage
path: root/STANDARDS.txt
blob: ad0f4b58aa7e7c31bddfbf852ab93c49cc6d8e92 (plain)
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
In a non experimental project, this document would describe the coding
standards used.  This is experimental code, we don't have any standards
yet.  Not even any standard language.  For the previous Java version of
matrix-RAD, I referenced a well known Java coding standard, and
documented the differences.  There's still some value in that old
standard, we can just ignore the Java specific bits.

In the end, we want to be language agnostic.  By using sockets / pipes /
stdin/out / etc, and the Unix philosophy of little tools you hook
together that each do one job well.  So we really don't care how badly
any given tool is written, so long as it works.  Some of them will be
written be strangers across the Internet.  This does need strong
communication standards though, which is what Nails is all about.

We do want to be secure though.  So eventually the non experimental
project this will grow into should have a core of well written code,
written to a standard.  Lacking any standard for this experimental
version, this document will just have some random notes about style and
standards.


Coding standards
----------------

I want to use C and Lua, with assembler for speed critical parts.  Speed
is an important issue, and these three are the best in their class for
speed.  When used correctly, all three can also keep the code size down,
which is also important.  Obviously security will be important for the
non experimental version, so keep that in mind.

So far Lua as a prototyping language has been working well.  So I can
see us roughing code out in Lua to start with, then rewriting it in C if
needed, then assembler if more speed is needed.

On the other hand, we do want full functionality to be available to high
level scripting languages.  While Lua is the chosen language, there's
been some success translating to and from Lua which may help us be more
agnostic here.  There's others working on translating things to Lua,
based on the theory that since it's the fastest scripting language, that
might help speed up other scripting languages.

In particular, translating LSL into Lua has so for been working well.  
The parts that can be benchmarked so far are very fast indeed, making
SL and OS look like the crippled snails that they are.  Also, there's
some evidence that we might be able to translate Lua into ECMAScript to
run in web browsers, or run a Lua VM inside the ECMAScript VM.  
Allegedly, while the scripts run slower than LuaJIT can run them, they
are still quite fast compared to ECMAScript.  We need this to support
legacy web browsers, which decided long ago to standardise on
ECMAScript.  Alice loves her some JavaScript and node.js.

We should allow early experimental work to be written in any language as
well.  With the proviso that if it becomes a core part later, it should
get rewritten in a combination of Lua, C, and assembler.


Documentation standards
-----------------------

This project started life as design documents on a MediaWiki based web
site, using MediaWiki markup.  Those design documents where migrated to
the Drupal system that replaced the site.  At this point every one
wanted to use a different markup syntax.  lol

I solved this by installing a few markup systems, but the majority of it
got written as what Drupal refers to as "simple formatting", which
basically allows the use of certain HTML tags, plus a few niceties.

Eventually code started to get written by me, I put it on GitHub.  At
around the same time, I was busy transferring the Team Purple website
from individual web applications to SourceForge.  Team Purple are the
developers of the Imprudence and Kokua viewers.  Later SourceForge
turned evil, so I started looking for other project hosting options.  
GitHub has a wiki part, but they squeeze you into a tiny column in the
middle of the web page, and you can't do colours.  BitBucket (the other
major project hosting site used by Team Purple) seems to want to sell
you JIRA, so might also not be suitable.  In either case, they both
might turn evil as well.  Can't find a decent project hoster, sigh.

It looks like it helps to be generic, instead of constantly converting
between each of these markup formats each time we change hosting system.  
All of these markup systems get parsed by active code somewhere at
viewing time, to get translated into HTML+CSS+JavaScript that is sent to
your browser.  All of the documents are static.  Sure, matrix-RAD
included a worlds first active manual, you could test out matrix-RAD
commands right there in the docs, but we are going a different way this
time.  We are no longer living in the web browser, but replacing it with
a virtual world, that might happen to include web pages on prims, but
that's not our main thrust for UI stuff.

Sooooo, to avoid a lot of work, how about we just make the documentation
static HTML web pages, and be done with it?  Markup systems are just a
way to make HTML more palatable to ordinary users, but ordinary users
are not writing the design documentation and such.  They all get
translated to HTML, so let's just write raw HTML ourselves.  Keep it
simple, no fancy CSS or Javascript shit.  So we avoid markup, we avoid
having to translate from markup to markup, and we avoid the rendering
step of translating markup to HTML.  We avoid having to parse the
documentation at all, just send it to any old web browser.

In this way, we can just throw the basic documentation up on any old web
server.  Enough for people to learn about it, get interested, and get it
up and running.  People editing this stuff could just use any HTML
editor if they like, including translating their favourite markup to HTML.

Currently there's ordinary text files as well.  A few of those should
stay as ordinary text, but some (or most) could get converted to HTML,
then linked in to the rest.  Also, I should link up the existing pages
now, some more internal navigation.


On the other hand, once they have the system running, there's -

Internal documentation standards
--------------------------------

matrix-RAD included plans for being able to edit down to the code level
from within the web browser running matrix-RAD apps.  SL and OS include
world building tools you use from inside the world, including editing of
scripts.  So SledjHamr is no different, it will include tools you can
use from inside the world, to edit everything about the world.  Which
should include stuff like the matrix-RAD JavaDocs extensions to help
self document code and other fancy things.  Obviously this will include
coding standards, and tools to help with those standards.

Once things start being less experimental, then we can decide how that
all works, and the coding standards to suit.


Communication standards
-----------------------

That's what Nails is all about, so see the docs for that.