aboutsummaryrefslogtreecommitdiffstatshomepage
path: root/src/NOTES.txt
blob: 1758ac80cd089b9048bd1e846af44c3f5343c180 (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
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
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
I'm re-purposing this for SledjHamr https://sledjhamr.org/git/docs/index.html

The general structure of SledjHamr is a bunch of servers talking to each
other via Internet (or just local) connections.  One of them is a web
server for assets, world data, and inventory. 

Originally I didn't think using a web based world client was a good idea,
however it might be better to have one, for reasons.  Now I need a web
management console that can do all the things the current tmux console
can, including OpenSim console and commands.  Plus account management for
users.  I can also use a web based Jabber / XMPP front end to chat, IM,
and group chatter, which would run in the normal viewers web browser. 
This provides a doorway into putting SledjHamr stuff in existing viewers
without needing them to support it.  So a web based viewer now makes more
sense, and also means we can get away with not needing a viewer at all.

Toybox itself doesn't include a web server, and I don't think there is
one on the roadmap.  So we have to use an external web server, which was
a design goal of SledjHamr in the first place, using existing mature
HTTP infrastructure, coz that's already solved problems for a bunch of
things that plague OS/SL to this day.  Clear your cache!  Pffft.

So sledjchisl.c will be the "love world server", though initially it just
drives OpenSim_SC in tmux via tmux commands to send keys and read output. 
Later it might run opensim_SC directly and use STDIN and STDOUT to do
everything.  It'll also provide the text management front end that runs
in the left tmux panel of the first window, which is why it's based on
boxes in the first place.  Later still it can take over opensim_SC
functions as I move them out of mono.

We will need a text, web, and GUI version of this management front end. 
Hmmm, maybe don't need a GUI version, GUI users can just run a terminal.

After much research, FastCGI / FCGI seems to be the most portable way of
interfacing with existing web servers.  FCGI protocol closes STDERR and
STDOUT, and uses STDIN as it's two way communications channel to the web
server, so our FCGI module can't be used as the text management front
end.  This is probably a good idea to keep them seperate anyway, for
security, coz the web server is exposed to the world, the console isn't.

Currently sledjchisl.c tests to see if it's running in tmux already, if
it isn't it starts up tmux runs itself into this new tmux, then exits. 
So it could also test if it's running from FCGI, and switch to web mode,
then it'll need to find the tmuxed instance to send commands to it. 
Either via nails connection, or sending tmux commands via shell.

FCGI has methods of dealing with auth and templates.  B-)

So for now I think I'll have the text and web management front ends in
sledjchisl.c, and the love world server as well.  I can split them up
later if I need to.




I has	Apache 2.4.25-3+deb9u9
	MariaDB 10.1.44-MariaDB


https://gist.github.com/dermesser/e2f9b66457ae19ebd116
    Multithreaded example in C.


-------------------------------------------------------------------

Apache doesn't seem to support FCGI filter role, so I might have to do
without.  Might be better anyway.


"A Filter is similar in functionality to a Responder that takes a data
file as a parameter. The difference is that with a Filter, both the data
file and the Filter itself can be access controlled using the Web
server's access control mechanisms, while a Responder that takes the name
of a data file as a parameter must perform its own access control checks
on the data file."

    Which is fine, our access control checks will be "Is this database
    defined user already logged on via our FCGI script?".  We should have
    total control over that.  I was planning on using the FCGI auth
    mechanism anyway.


RESPONDER
    web server sends FCGI_PARAMS
	CONTENT_LENGTH
    web server sends input body FCGI_STDIN
    fcgi app sends result data over FCGI_STDOUT and error messages over FCGI_STDERR
	it has to finish reading FCGI_PARAMS first
    fcgi app sends FCGI_END_REQUEST(protocolStatus = FCGI_REQUEST_COMPLETE)


FILTER
    filtered file has last modified time
	web server sets FCGI_DATA_LAST_MOD accordingly
    web server sends FCGI_PARAMS
	CONTENT_LENGTH   FCGI_DATA_LAST_MOD   FCGI_DATA_LENGTH
    web server sends input body FCGI_STDIN
    web servers sends file over FCGI_DATA
    fcgi app can ignore FCGI_DATA and use it's own cached copy based on FCGI_DATA_LAST_MOD
    fcgi app sends result data over FCGI_STDOUT and error messages over FCGI_STDERR
	it has to finish reading FCGI_STDIN first, but not FCGI_DATA
    fcgi app sends FCGI_END_REQUEST(protocolStatus = FCGI_REQUEST_COMPLETE)


Soooo, FILTER might be slower anyway if we are caching the filtered file,
or mmapping it, coz filter has to start sending the filtered file, even
if it's to be served from cache.  Plus no need to wait for FCGI_STDIN
before spewing it out.


PLAN -
.    add "webRoot" to the config, point it at /opt/opensim_SC/web
.	/opt/opensim_SC/web/fcgi-bin/sledjchisl.fcgi	a symlink to /opt/opensim_SC/bin/sledjchisl
.	/opt/opensim_SC/web/html/			the static web files and templates.
.    https://localhost/opensim_SC/fcgi-bin/sledjchisl.fcgi/foo.html
.	check the if modified since bit against the replaceable bits last fetch time / /opt/opensim_SC/web/html/foo.html last modified time
.	    check if /opt/opensim_SC/web/html/foo.html is cached, or if it's been modified since last cache.
.		mmap /opt/opensim_SC/web/html/foo.html
.		put it in a cache
.		scan through it looking for the replacable bits
.		    store pointers to them
.	    check if the replaceable bits need refreshing
.	    loop through the pointers to the cache
.		spew out the non replacable bits
.		spew out the replacements for the replacable bits
.		repeat until done

Last update time for parameters, plus an update frequency.  Once a minute.
Hash of page file names
.    last modified time for file
    Linked list of page fragments  ->  array (qlibc can convert a linked list into an array?), or try the growable thingy.
.	struct
.	    enum telling us what this bit is
.		bit of verbatim text
.		replaceable parameter, pointer to the data that is also stored in the ssi hash, so that changes propogate
.		    <!--#echo var="???" -->
		bit of Lua

    NOTE - SSI is a bit more complex than what I'm currently using.
	https://en.wikipedia.org/wiki/Server_Side_Includes
	    <!--#include virtual="menu.cgi" -->
	    <!--#include file="footer.html" -->
	    <!--#exec cgi="/cgi-bin/foo.cgi" -->
	    <!--#exec cmd="ls -l" -->
.	    <!--#echo var="REMOTE_ADDR" -->
	    <!--#config timefmt="%y %m %d" -->
	    <!--#config sizefmt="bytes" -->
	    <!--#config errmsg="SSI command failed!" -->
	    <!--#flastmod virtual="index.html" -->
	    <!--#fsize file="script.pl" -->
	    <!--#if expr="${Sec_Nav}" -->
	    <!--#include virtual="secondary_nav.txt" -->
	    <!--#elif expr="${Pri_Nav}" -->
	    <!--#include virtual="primary_nav.txt" -->
	    <!--#else -->
	    <!--#include virtual="article.txt" -->
	    <!--#endif -->
	    <!--#set var="foo" value="bar" -->
	    <!--#printenv -->
	https://www.w3.org/Jigsaw/Doc/User/SSI.html
	    Adds lots of others, including Java stuff.
	Mine
	    <!--#lua lua="print(table[key])" -->
	    <!--#lua file="/path/to/script.lua" -->
	    <!--#lua virtual="https://example.com/script.lua" -->

BTW - /opt/opensim_SC/web/html/foo.html should have it's image URLS and other
static data set to https://localhost/opensim_SC/SledjHamr.png, which
would map to /opt/opensim_SC/web/html/SledjHamr.png and be treated as ordinary
static files by the web server.

ALSO - when spewing results, we have to manually send the headers first
our selves, this should include the HTTP status code and string, content
type, cookies, etc.

-------------------------------------------------------------------

https://project-awesome.org/aleksandar-todorovic/awesome-c
    A curated list of C good stuff.

https://wolkykim.github.io/qdecoder/
    CGI library made by the qlibc guy, does support FCGI.
    Might be a wrapper around the fcgi_stdio stuff I'm already using?