Re: [sqlite] new sqlite-based webserver

Günter Greschenz Thu, 12 Oct 2006 01:04:53 -0700

hi raymond,

your idea of including other files is already done.

there is a function "script" (see "gas,js") which includes other files.that how i want to support different "applications". on example is"notes". in "apps/notes.js" i implemented the ajax backend, in"www/Notes/*" i implemented the frontend.

the part you've written about loading xml files (contacts.xml) is solvedlike this (again, see gas.js):

   var config = new XML(new File("server.xml").read());
   print(config.port+"\n");

a simple scheduler is also already implemented (again, see gas,js :-):
   // print "from timer" to console in 10 secs
   addTimer(10, "print('from timer\\n');");

i also implemented a database sample for directly editing databases viathe browser (see SQLiteBrowser in the source zip). but its still notfinished (only running in MSIE correctly !)there you can play with databases: select a database / select a table /edit values just with a dblclick, change the value, hit return...


cu, gg


Raymond Irving schrieb:

This is very very sweet :)

Many thanks for sharing Greschenz. I would love to see
this project mature. Maybe something like a ThinClient
webserver with Javascript as the backend so we could
write lovely web apps with AJAX support that offers
database like features? Could it also run on

Thinstation with some attached storage? Wow!

Here's some other ideas:

* Make it possible to include external classes or
files, for example include('myclass.jsp')
* add some build it object to make it easier to do
stuff inside the server. Example Mail, File, Scheduler

var mail = new Mail()
mail.to = '[EMAIL PROTECTED]';
mail.from = '[EMAIL PROTECTED]'
mail.subject = 'Hello';
mail.send()

var f = new File('contacts.xml');
var xml = f.readAll();

var sch = new Scheduler();
if(!sch.exists('MyImporter')) {
   sch.setName('MyImporter');
   sch.setInterval(30); // every 30 minutes
   sch.execute('imports.jsp');
   sch.save();
}


Just my 2 cents.
__
Raymond Irving


__
Raymond Irving

--- Günter Greschenz <[EMAIL PROTECTED]> wrote:

Hi,

the reasons why i did this:
1) i wanted to have a simple webserver to get my son

involved inprogramming an online game, and he knows already a

little bit javascript
2) my hobby is programming in c++ and javascript,

but at work i have tocode in c#, so i just wanted to have some fun :-)


i think its no a new idea, ive seen some webservers

using javascript asbackend language, but everything i found was tocomplicated toinstall/use or to complicated to port to a newhw-platform (i want torun it on my linksys-nslu2 with ftpd to my topfieldhd-vcr to programthe recordings via web).

and i like sqlite very much and no webserver i found
had this combination.
if you look at my code, you see that i just use
*) sqlite
*) the javascript-engine from mozilla (very

advanced: js v1.7 includingxml support...),

*) some glue code to access sqlite and a simple http

server from js(only one file: gas.cpp)


the trick between js and sqlite is like yours: every

sql statementexecuted returns a 2-dim array...

one thing is maybe iteresting: the class "DBItem".

its a kind of (verysimple) or-mapper.

it maps the properties of a javascript-object to the

columns of a table.here is a code example:


    var db = new Database("user.sdb");
    db.exec("create table if not exists user (id

integer primary keyautoincrement, name varchar unique, pwd varchar,

test varchar)");
    var user = new DBItem(db, "user");
    user._name = "Guenter";
    user._pwd = "FooBar1234567890";
    user._test = "blubb";
    user.flush();
    var id = user._id;
    var user = new DBItem(db, "user", id);
    print("name="+user._name+"\n");
    //db.exec("delete from user where id="+id);

if you are still curious (or anyone else ?), i can

send you the actualsource by mail. :-)


btw: i had a look into your http-source: looks nice,

but its maybe tocomplicated to implement my features like

http-multipart-posts.

cu, gg





John Stanton schrieb:

The multi-threaded application/www server I

described requires no

threading involvement from the application

programmer. That

programmer uses SQL, HTML, Javascript and the

application language we

call MUV. Think about it, when you use Apache you

don't have to be

aware of its internal threading.

The only significant synchronisation element is to

do with multiple

users of Sqlite, and that is handled transparently

to the application

programmer.

What I was curious about was your statement that

you used Javascript

as a backend. Do you have a novel idea there?

For example my

application language will, inter alia, populate

Javascript arrays from

an SQL statement.  Do you have any such features

or something more

ingenious? (I still cannot get access to your

server).

My application server runs on Win98, Win2000/XP,

Linux, AIX 4.2 and

5.3.  The executive, fileserving, compression and

CGI components are

realised in less than 20K of code.  HTML V2 is

implemented.

There are some old fragments of this program athttp://www.viacognis.com/muvm. If you are

interested I can dig out

the current code and let you have it.


Günter Greschenz wrote:

Hi,

normally i like multithreading, but in this case

i like the simple

approach (at least for rendering the pages),

because as application

programmer (the guy who will develop the apps in

javascript) its much

easier if you dont have to think about locks,

racing conditions and

other pitfalls. and the rendering itself is

(depending on what you

do, of course) fast enough for me (simple pages

need <10ms !). I want

to multithread the upload to the client after all

work is done.

in my current implementation the complete server

is blocked until the

download is finished. in my logfiles i have

entries with 25 minutes

(see

"http://greschenz.dyndns.org/logsByTime.html";) !!!

thats really

unacceptable !
could you mail me some samples howto multithread

portable (linux &

win32) ?

cu, gg


John Stanton schrieb:

Gunter,

I recently wrote a multi-threaded portable web

server in simpl ANSI

C. It uses Win32 threads or pthreads.  It caches

threads and re-uses

them on a most recently used basis.  Efficiency

is obtained by using

TransmitFile (Win32) or sendfile (Unix).

The logic is simple for an efficient,

multi-threaded www server.

An added bonus of mutli-threading is that it

takes advantage of the

ability of a browser to open multiple concurrent

connections and

persistent connections.

My WWW server is set up as an application

server, with embedded

Sqlite and an embedded byte-coded metalanguage

used to define DHTML

pages.  An associated compiler produces the byte

code. By using

compiled byte code the overhead of interpreting

a script-type

language is avoided.

Günter Greschenz wrote:

Hi,

yes i know, the problem is, its single-threaded

(because of

protability), so if anyone starts a download

with a slow

connection, the server is blocked for other

connections.. im still

thinking about this problem... single-threading

whle rendering th

pages and multithreading for sending them to

the clients, or maybe

async socket writes (is this possible in linux

?)

but i've seen (in the logs :-) a lot of people

are interested. its

funny to sit at the console at home and see the

log messages when

anyone comes by...
at the moment its just an alpha-version... but

if i have more time

(or maybe anyone helps me) it will improve !
the server itself is an i386-linux at ~ 300mhz,

so dont expect too

=== message truncated ===


-----------------------------------------------------------------------------
To unsubscribe, send email to [EMAIL PROTECTED]
-----------------------------------------------------------------------------

Re: [sqlite] new sqlite-based webserver

Reply via email to