[9fans] httpd scripting

[9fans] httpd scripting

[Skip Tavakkolian]

Has anyone done any deep thinking on this? I've just started
looking into it. I have looked at Pegasus, but I need something a
little more specialized.  My own preference is for an awk-like environment
but nothing definite yet. I want to process code fragments inside
html docs, the same thing as jsp,asp,php, etc.

P.S.
Yes, I could go with Apache or something else on a different
OS, but I don't want to.

Re: [9fans] httpd scripting

[Kenji Arisawa]

Hello,

 >Has anyone done any deep thinking on this? I've just started
 >looking into it. I have looked at Pegasus, but I need something a
 >little more specialized.  My own preference is for an awk-like
environment
 >but nothing definite yet. I want to process code fragments inside
 >html docs, the same thing as jsp,asp,php, etc.

I think the solution is
(html scanner) + (bidirectional pipe to a program)
though Pegasus does not offer such a scanner yet.

Sorry if I misunderstand what you say.

Kenji Arisawa

Re: [9fans] httpd scripting

[Skip Tavakkolian]

> I think the solution is
> (html scanner) + (bidirectional pipe to a program)
> though Pegasus does not offer such a scanner yet.

I like that.

Also, I am (sort of) considering a fs for scripting, mounted
in httpd's namespace; Not sure how the user's client context
would be forwarded to the fs (cookies, and the like).

Re: [9fans] httpd scripting

[Skip Tavakkolian]

>> .... (cookies, and the like).
>
> those things should be blocked.

The only legitimate use I can think of (for a server) is keeping
track of a session (identifying a session).

Re: [9fans] httpd scripting

[boyd, rounin]

> .... (cookies, and the like).

those things should be blocked.

Re: [9fans] httpd scripting

[boyd, rounin]

> The only legitimate use I can think of ....

i don't think it's 'legitimate use'

Re: [9fans] httpd scripting

[Kenji Arisawa]

Hello,

 >Also, I am (sort of) considering a fs for scripting, mounted
 >in httpd's namespace;

Fs idea may be interesting but might bring other problems.

Kenji Arisawa

Re: [9fans] httpd scripting

[matt]

Skip Tavakkolian wrote:

>Has anyone done any deep thinking on this? I've just started
>looking into it. I have looked at Pegasus, but I need something a
>little more specialized.  My own preference is for an awk-like environment
>but nothing definite yet. I want to process code fragments inside
>html docs, the same thing as jsp,asp,php, etc.
>

as someone who uses PHP (and previously ASP) professionally - don't 8)

RC and friends is enough

if want awk then use awk  8)

as for cookies, headers etc. you just need a quick script to chuck the
request into the environment (which is what PHP does)

I did write a script to do it once

http://groups.google.com/groups?hl=en&lr=&ie=UTF-8&oe=UTF-8&safe=off&threadm=015d01c22ed3%24fcc596e0%246501a8c0%40KIKE&rnum=1&prev=/groups%3Fq%3Dgroup:comp.os.plan9%2Bhttpd%2Bmatt%26hl%3Den%26lr%3D%26ie%3DUTF-8%26oe%3DUTF-8%26safe%3Doff%26selm%3D015d01c22ed3%2524fcc596e0%25246501a8c0%2540KIKE%26rnum%3D1

but I seem to have lost the code during a re-install (curses for putting
the magic directory outside of /usr )

All you really need to do is split the headers on the first colon :
([^:]+): ?(.*)

(after skipping past the first line)
and then pop the results into the environment

echo $match_2 > /env/$match_1

when you run out of matches the rest of stdin is the body & you might
even be able to trust the Content-Length value

Re: [9fans] httpd scripting

[David Butler]

I did a lot of work in the http area on Plan 9, back on
version 2 of the system. I took the approach that anything
is executable (compiled) instead of scripting. It was fast
and very flexible. Perhaps I can see what it takes to get
it on version 4 and make it available...

David

----- Original Message -----
From: "matt" <matt@proweb.co.uk>
To: <9fans@cse.psu.edu>
Sent: Monday, June 02, 2003 6:55 AM
Subject: Re: [9fans] httpd scripting


> Skip Tavakkolian wrote:
>
> >Has anyone done any deep thinking on this? I've just started
> >looking into it. I have looked at Pegasus, but I need something a
> >little more specialized.  My own preference is for an awk-like
environment
> >but nothing definite yet. I want to process code fragments inside
> >html docs, the same thing as jsp,asp,php, etc.
> >
>
> as someone who uses PHP (and previously ASP) professionally - don't 8)
>
> RC and friends is enough
>
> if want awk then use awk  8)
>
> as for cookies, headers etc. you just need a quick script to chuck the
> request into the environment (which is what PHP does)
>
> I did write a script to do it once
>
>
http://groups.google.com/groups?hl=en&lr=&ie=UTF-8&oe=UTF-8&safe=off&threadm=015d01c22ed3%24fcc596e0%246501a8c0%40KIKE&rnum=1&prev=/groups%3Fq%3Dgroup:comp.os.plan9%2Bhttpd%2Bmatt%26hl%3Den%26lr%3D%26ie%3DUTF-8%26oe%3DUTF-8%26safe%3Doff%26selm%3D015d01c22ed3%2524fcc596e0%25246501a8c0%2540KIKE%26rnum%3D1
>
> but I seem to have lost the code during a re-install (curses for putting
> the magic directory outside of /usr )
>
> All you really need to do is split the headers on the first colon :
> ([^:]+): ?(.*)
>
> (after skipping past the first line)
> and then pop the results into the environment
>
> echo $match_2 > /env/$match_1
>
> when you run out of matches the rest of stdin is the body & you might
> even be able to trust the Content-Length value
>
>
>
>
>

Re: [9fans] httpd scripting

[Charles Forsyth]

>>RC and friends is enough

i did found that sufficient when i did
a plan 9 httpd years ago, probably before asp let alone php
had appeared.  i used echo or cat and here documents quite a bit.
i still think it seems better to generate the html
within the (obviously) shared dynamic environment provided by
the rc script rather than have lots of executable fragments embedded in html.
it needn't be rc of course.

Re: [9fans] httpd scripting

[matt]

Charles Forsyth wrote:

>>>RC and friends is enough
>>>
>>>
>
>i did found that sufficient when i did
>a plan 9 httpd years ago, probably before asp let alone php
>had appeared.  i used echo or cat and here documents quite a bit.
>i still think it seems better to generate the html
>within the (obviously) shared dynamic environment provided by
>the rc script rather than have lots of executable fragments embedded in html.
>it needn't be rc of course.
>

the "code fragments in the HTML" model is not very scalable

in fact, content management becomes a total nightmare

you have to move from
"how is the HTML going to display the results of executing this code"
to
"what HTML should the code produce"

It's like it starts you off at the wrong layer of abstraction

Re: [9fans] httpd scripting

[Charles Forsyth]

i looked at the code, and the rule i used was simply
that if the url for GET and POST corresponded to an executable file--
as gdb suggested not just an rc file, though i did use rc for many examples--
httpd forked the name space (having originally secured it)
and environment, put some of the parsed data in the environment
(could have served it up in files), optionally sent an HTTP
reply header, and ran the program.

if the executable file's suffix was .http, it was assumed to
generate its own HTTP headers, otherwise httpd provided whatever
was appropriate to the file's suffix.

Re: [9fans] httpd scripting

[Ian Broster]

> the "code fragments in the HTML" model is not very scalable
>
> in fact, content management becomes a total nightmare

I've come to the conclusion that neither 'code in HTML' nor 'HTML
in code' are very good! Neither are easy to read, quotes mess
everything up and it's hard to debug.

My thought is to try to raise the abstraction higher.
Here's how I'd like to code a server generated page.

int some_data[5][5]; /*eg*/
create_my data(some_data);
...
{
 new_web_page(template);
 add(some_data, how_to_add_intgers);
 add(sometext, how_to_add_text);
 ...
 done(); /* composes the page and writes it out*/
}


Note that the ordering is not 'linear'. It doesn't matter which
order the add() calls come. the 'how_to_add_text' things
(whatever they are?) describe how the thing is presented in
the markup language.

Perhaps this has overtones of a Tk-like interface.
Perhaps it's more object oriented. One
day, I'd like to try to find the time to investigate.
Maybe something already exists?

ian

Re: [9fans] httpd scripting

[matt]

>
>
>My thought is to try to raise the abstraction higher.
>Here's how I'd like to code a server generated page.
>
>int some_data[5][5]; /*eg*/
>create_my data(some_data);
>...
>{
> new_web_page(template);
> add(some_data, how_to_add_intgers);
> add(sometext, how_to_add_text);
> ...
> done(); /* composes the page and writes it out*/
>}
>
>
that's kind of how I do it
<?
$this->add_head_item('<link href="/cascading_menu.css" rel="stylesheet"
type="text/css">');
$this->add_head_item('<link href="/cascading_menu_colours.css"
rel="stylesheet" type="text/css">');
$this->add_h1('useful links');
$this->add_anchor('plan 9 homepage',
array('href'=>'http://plan9.bell-labs.com/plan9', 'target'=>'_blank');

echo $this->get();
?>

and out  pops a valid, well formatted html document

I've added a few things to it over that last 6 months such that this site

http://www.bigchoicemobilephones.com

took about a week to build from scratch

the body content sits in a directory tree outside of the webspace & the
php pulls it all together

Re: [9fans] httpd scripting

[spam]

> Have you seen tclhtppd?  It's on sourceforge.

Have now. Thanks.

Seems to be quite well thought out, although it is basically 'code in HTML', the separation of a template file from the code
seems sensible.


ian

Re: [9fans] httpd scripting

[matt]

David Butler wrote:

>I did a lot of work in the http area on Plan 9, back on
>version 2 of the system. I took the approach that anything
>is executable (compiled) instead of scripting. It was fast
>and very flexible. Perhaps I can see what it takes to get
>it on version 4 and make it available...
>
>David
>

my only beef with ip/httpd is that it maps URIs to filenames
(not that it keeps me awake at night)

I've got over 500,000 URIs on one of my sites and, as such, mapping
them, even via namespaces would be the wrong approach

the world is an easier place if

http://www.thebigchoice.com/Graduate_Jobs/IT_and_Management_Systems/York

is the URI instead of

http://www.thebigchoice.com/magic/show_jobs?section=Graduate_Jobs&job_section=IT_and_Management_Systems&area=York

esp. when you factor in relative URIs in the resulting HTML

m

Re: [9fans] httpd scripting

[Joel Salomon]

On Mon, 2 Jun 2003, Ian Broster wrote:
>
> Perhaps this has overtones of a Tk-like interface.
> Perhaps it's more object oriented. One
> day, I'd like to try to find the time to investigate.
> Maybe something already exists?
>
> ian
>
This feels similar to XML+DOM, which may be too buzzword-compliant for
plan9, but was invented for this problem domain.

--Joel

Re: [9fans] httpd scripting

[Robby]

In article <20030602163818.316ebbf5.spam@broster.co.uk>,
	spam@broster.co.uk (Ian Broster) writes:
>> the "code fragments in the HTML" model is not very scalable
>>
>> in fact, content management becomes a total nightmare
>
> I've come to the conclusion that neither 'code in HTML' nor 'HTML
> in code' are very good! Neither are easy to read, quotes mess
> everything up and it's hard to debug.
>
> My thought is to try to raise the abstraction higher.
> Here's how I'd like to code a server generated page.
>
> int some_data[5][5]; /*eg*/
> create_my data(some_data);
> ...
> {
>  new_web_page(template);
>  add(some_data, how_to_add_intgers);
>  add(sometext, how_to_add_text);
>  ...
>  done(); /* composes the page and writes it out*/
> }
>
>
> Note that the ordering is not 'linear'. It doesn't matter which
> order the add() calls come. the 'how_to_add_text' things
> (whatever they are?) describe how the thing is presented in
> the markup language.
>
> Perhaps this has overtones of a Tk-like interface.
> Perhaps it's more object oriented. One
> day, I'd like to try to find the time to investigate.
> Maybe something already exists?
>
> ian

This kind of reminds me of a posting to comp.compilers, which I love to whip
out whenever somebody wants to understand function pointers or variable
arguments. I don't know if this "just" works in plan9, though. I haven't
had the need to try something like this out yet.

Robby


> From telnet@wagner.Princeton.EDU.composers Wed Jul  8 06:40:19 1998
> Newsgroups: comp.compilers
> Subject: Re: writing an assembler!
> Date: 8 Jul 1998 01:40:01 -0400
> Organization: Chemistry Department, Princeton University
> Message-ID: <98-07-062@comp.compilers>
> Keywords: macros, design
> Lines: 70

> In article <98-07-005@comp.compilers>, "Dr Richard A. O'Keefe" <ok@atlas.otago.ac.nz> writes:
> > These days, my language of choice for doing this would be Lisp or Ada
> > (sensible macros are about *trees*, not strings!) but you can
> > certainly use C as an assembler quite effectively.
> > [It's a disgusting hack, but it's a great disgusting hack. -John]
>
> You can get away with tree-like behavior in C if you use varargs stuff.
> I'll risk my reputation (:-)) by admitting to an even more disgusting hack:
> ---
> void new_section(char *title, char *img_name, int h, int w) {
>     emit(html_list,
> 	    html_p,
> 	    html_hr,
> 	    html_br,
>             html_region, "center",
>                 html_region, "h1",
> 	            html_text, title,
>             html_tag, "img",
> 	        html_param, "src", html_string, fname,
> 	        html_param, "height", html_text, h,
> 	        html_param, "width", html_text, w,
> 	        0,
>             0);
> }
> ---
> typedef void (*html_func)(va_list *);
>
> void emit(html_func first, ...) {
>     va_list ap;
>
>     va_start(ap, first);
>     (*first)(&ap);
>     va_end(ap);
> }
>
> void html_list(va_list *ap) {
>     html_func f;
>
>     while (1) {
> 	f = va_arg(ap, html_func);
> 	if (f == 0) break;
>
> 	(*f)(ap);
>     }
> }
>
> [etc]
> ---
>
> A disgusting hack?  Absolutely.
>
> But it's also an order of magnitude more powerful than any other
> dynamic HTML tool I've worked with, and it only took me an hour or so
> to implement.
> ---------------------------------------------------------------------------
> Tim Hollebeek
> email: tim@wfn-shop.princeton.edu
> URL: http://wfn-shop.princeton.edu/~tim
>
> [Wow, that's loathsome.  I really like it. -John]

Re: [9fans] httpd scripting

[William Ahern]

 From telnet@wagner.Princeton.EDU.composers Wed Jul  8 06:40:19 1998
Newsgroups: comp.compilers
> You can get away with tree-like behavior in C if you use varargs stuff.
> I'll risk my reputation (:-)) by admitting to an even more disgusting hack:
> ---
> void new_section(char *title, char *img_name, int h, int w) {
>     emit(html_list,
> 	    html_p,
> 	    html_hr,
> 	    html_br,
>             html_region, "center",
>                 html_region, "h1",
> 	            html_text, title,
>             html_tag, "img",
> 	        html_param, "src", html_string, fname,
> 	        html_param, "height", html_text, h,
> 	        html_param, "width", html_text, w,
> 	        0,
>             0);
> }
<snip>
>
> A disgusting hack?  Absolutely.
>
> But it's also an order of magnitude more powerful than any other
> dynamic HTML tool I've worked with, and it only took me an hour or so
> to implement.
> ---------------------------------------------------------------------------

this is very similar to XSL: trees, nodes, concept of tags built into
language. c/o xsltproc from libxml2 for an XSLT processor, and zvon.org
for really good tutorials.

the beauty of XSL is that you can cleanly separate your content production
logic from your presentation logic.

- Bill