pop3-dev@httpd.apache.org Archives

[yoshiki@xemacs.org: Re: Docs for mod_pop3]

2003-12-10T19:20:51Z

http://www.apacheworld.org/pop3/

I wrote some docs for mod_pop3, in the standard Apache format, does anyone
want to review/commit?

Best regards

Daniel
----- Forwarded message from Yoshiki Hayashi  -----

Delivered-To: daniel@rawbyte.com
Mailing-List: contact docs-help@httpd.apache.org; run by ezmlm
Precedence: bulk
list-help: 
list-unsubscribe: 
list-post: 
Reply-To: docs@httpd.apache.org
Delivered-To: mailing list docs@httpd.apache.org
Mail-Copies-To: nobody
To: docs@httpd.apache.org
Subject: Re: Docs for mod_pop3
From: Yoshiki Hayashi 
Date: Tue, 09 Dec 2003 17:36:08 -0700
In-Reply-To: <20031207213716.GA15109@rawbyte.com>
 (Daniel Lopez's message of "Sun, 7 Dec 2003 13:37:16 -0800")
User-Agent: T-gnus/6.15.16 (based on Oort Gnus v0.16)
X-Spam-Rating: daedalus.apache.org 1.6.2 0/1000/N
X-Spam-Status: No, hits=-6.3 required=5.0
	tests=BAYES_01,EMAIL_ATTRIBUTION,IN_REP_TO,REFERENCES,
	      REPLY_WITH_QUOTES,USER_AGENT
	version=2.55
X-Spam-Level: 
X-Spam-Checker-Version: SpamAssassin 2.55 (1.174.2.19-2003-05-19-exp)

Daniel Lopez  writes:

> I wrote some docs for mod_pop3 back in the days, can somebody have a look
> and review/include them?
>
> http://www.apacheworld.org/pop3/

Wow, great!

I really like to see it in the docs but I think it should go
to httpd-pop3 repository instead of httpd judging by the
previous experience with mod_ldap.  (When mod_ldap was moved
out from httpd, documentation was also migrated.  When it
was added back to httpd, documentation was also added back.)

Perhaps you should try pop3-dev@httpd so that it can be
included in the distribution.

-- 
Yoshiki Hayashi

---------------------------------------------------------------------
To unsubscribe, e-mail: docs-unsubscribe@httpd.apache.org
For additional commands, e-mail: docs-help@httpd.apache.org

----- End forwarded message -----

-- 
Teach Yourself Apache 2 -- http://apacheworld.org/ty24/

Gifts for you

2003-10-20T13:10:33Z




Free Games





  
   
    
  
   
     
  
  
    This 
        mail is from your friends at SearchMusic.info.

        To stop receiving special offers, coupons, and discounts, click

        here

Gifts from Avon

2003-10-20T06:20:13Z




Free Games





  
   
    
  
   
     
  
  
    This 
        mail is from your friends at LoansApproved.biz.

        To stop receiving special offers, coupons, and discounts, click

        here

Gifts from Avon

2003-10-19T16:45:07Z




Free Games





  
   
    
  
   
     
  
  
    This 
        mail is from your friends at Sure-games.

        To stop receiving special offers, coupons, and discounts, click

        here

Daily Photo Gallery

2003-10-19T05:32:26Z




Free Games





  
   
    
  
   
     
  
  
    This 
        mail is from your friends at Myfoto.us.

        To stop receiving special offers, coupons, and discounts, click

        here

Photos for you

2003-10-18T22:30:45Z




Free Games





  
   
    
  
   
     
  
  
    This 
        mail is from your friends at Myfoto.us.

        To stop receiving special offers, coupons, and discounts, click

        here

Photos for you

2003-10-18T21:49:58Z




Free Games





  
   
    
  
   
     
  
  
    This 
        mail is from your friends at Myfoto.us.

        To stop receiving special offers, coupons, and discounts, click

        here

Gifts for you

2003-10-18T18:09:32Z




Free Games





  
   
    
  
   
     
  
  
    This 
        mail is from your friends at Sure-Games.com.

        To stop receiving special offers, coupons, and discounts, click

        here

RE: [PATCH] Logging, Two dots, MMAP, etc

2003-10-07T17:44:34Z

At 06:21 PM 10/6/2003, Sung Kim wrote:

>PS - Can I assume that the previous patch for http_pop3 is rejected? :)

Please don't assume that.  The pop3 contributors have alot of cycles on other
projects or security aspects of the httpd project and are sometimes overtaxed.

I'm looking forward to looking at your submission when my P1 queue empties
out again, in the meantime, I hope another committer has some cycles to look
and comment/correct/commit your proposal.

Looks like I need to edit my filters to split this list out of my dev@httpd box,
so it's easier to keep track of your efforts and other's comments :-)

Bill

RE: [PATCH] Logging, Two dots, MMAP, etc

2003-10-07T05:49:22Z

On Mon, 6 Oct 2003, Sung Kim wrote:

> PS - Can I assume that the previous patch for http_pop3 is rejected? :)

No, I just haven't gotten around to committing it yet.  Lots to do, little
time to do it in... :)

If somebody else has a minute, feel free to jump in...

--Cliff

RE: [PATCH] Logging, Two dots, MMAP, etc

2003-10-06T23:21:31Z

Hello pop3 developers,

We were working on mod_smtp that is a simple SMTP protocol handler using
Apache2 server.
We need mod_pop3 to test mod_smtp. Mod_smtp is getting messages and mod_pop3
transfer mail to MUA.
It means we are testing mod_pop3 a lot!

When I try to retrieve a mail using RETR command, I've got this error. It
seems the error is from deep in somewhere protocol.c.

Do you have any idea?

PS - Can I assume that the previous patch for http_pop3 is rejected? :)
- Sung

(gdb) r -X
Starting program: /usr/local/apache2/bin/httpd -X
[New Thread 8192 (LWP 14150)]

Program received signal SIGSEGV, Segmentation fault.
[Switching to Thread 8192 (LWP 14150)]
0x40019828 in apr_brigade_write (b=0xa, flush=0x80981b8 ,
    ctx=0x816a338, str=0x80a9a1b "+OK ", nbyte=4) at apr_brigade.c:406
406         apr_bucket *e = APR_BRIGADE_LAST(b);
(gdb) where
#0  0x40019828 in apr_brigade_write (b=0xa, flush=0x80981b8
,
    ctx=0x816a338, str=0x80a9a1b "+OK ", nbyte=4) at apr_brigade.c:406
#1  0x0809a387 in buffer_output (r=0x81799a0, str=0x80a9a1b "+OK ", len=4)
    at protocol.c:1369
#2  0x0809a463 in ap_rwrite (buf=0x80a9a1b, nbyte=4, r=0x817a360)
    at protocol.c:1404
#3  0x08067656 in generate_scan_listing (r=0x81799a0, mbox=0x817a688,
num=55,
    with_ok=1) at pop_protocol.c:222
#4  0x08067d1b in ap_handle_retr (r=0x81799a0, buffer=0x817a357 "")
    at pop_protocol.c:499
#5  0x080674be in process_pop_connection_internal (r=0x81799a0,
bb=0x8171a18)
    at pop_protocol.c:155
#6  0x08066f64 in process_pop_connection (c=0x8171a18) at pop_core.c:148
#7  0x08096106 in ap_run_process_connection (c=0x8169fb0) at connection.c:85
#8  0x0808bec3 in child_main (child_num_arg=135734136) at prefork.c:694
#9  0x0808c06e in make_child (s=0x810f148, slot=0) at prefork.c:734
#10 0x0808c0c7 in startup_children (number_to_start=5) at prefork.c:806
#11 0x0808c7b9 in ap_mpm_run (_pconf=0x808b7ac, plog=0x8109bc8, s=0x810f148)
    at prefork.c:1022
#12 0x080914d2 in main (argc=2, argv=0xbffffa14) at main.c:660
#13 0x420158d4 in __libc_start_main () from /lib/i686/libc.so.6

Re: [PATCH] Logging, Two dots, MMAP, etc

2003-09-28T01:09:55Z

Hello Cliff,

Thanks for the comments. I attached new patch based on your comments.
Also I'm not very sure about extra apr_rflush. But using the extra apr_rflush solves the two
dots problem.
For some reasons,  ap_rwrite(".\r\n", strlen(".\r\n"), r) produces two dots.

Have a nice day.

-
Sung Kim 
http://www.cse.ucsc.edu/~hunkim

 "Dreams become reality!"

On Sat, 27 Sep 2003, Cliff Woolley wrote:

> On Sat, 27 Sep 2003, Sung Kim wrote:
>
> > 1. Logging for each POP3 command and return status.
> > 2. Two dots from LIST and TOP commands.
> > 3. Error on apr_mmap_create. apr_mmap_create will return error for
> > APR_BURRERED file.
> > 4. Added file_close and apr_mmap_delete for cleanup.
>
>
> Hey, thanks for the patch!  A few minorcomments inline:
>
> > +++ pop_mbox.c	27 Sep 2003 06:50:40 -0000
> > @@ -65,7 +65,7 @@
> >  apr_status_t pop_parse_maildrop(request_rec *r, pop_mbox **mbox)
> >  {
> >      apr_status_t rv;
> > -    char *str = apr_pcalloc(r->pool, POP_STRING_LENGTH);
> > +    char str[POP_STRING_LENGTH];
> >      apr_off_t off = 0;
> >      int state = 0;
> >      int messages = 1;
>
> What's the purpose of this change?  Are you seeing a memory leak when
> pcalloc is used?  We tend to prefer that against stack buffers whenever
> possible.
>
> > @@ -83,7 +83,7 @@
> >
> >      ap_run_translate_name(r);
> >
> > -    if ((rv = apr_file_open(&ur->fp, r->filename, APR_READ | APR_WRITE
| APR_BUFFERED,
> > +    if ((rv = apr_file_open(&ur->fp, r->filename, APR_READ | APR_WRITE,
> >                         APR_OS_DEFAULT, ur->p)) != APR_SUCCESS) {
> >          ap_fprintf(r->output_filters, bb,
> >                     "-ERR, unable to open %s's maildrop.\n", ur->user);
>
> Good catch!
>
>
> > @@ -111,7 +111,7 @@
> >
> >          rv = apr_file_gets(str, bytes, ur->fp);
> >          if (rv != APR_SUCCESS && rv != APR_EOF) {
> > -            return rv;
> > +	    return rv;
> >          }
> >
> >          if ((!strncmp(str, "From ", strlen("From "))) || rv == APR_EOF) {
>
> Please try to avoid stylistic changes in functional patches; it makes them
> harder to review.  Also please refer to the coding styleguide at
> http://httpd.apache.org/dev/styleguide.html (in particular, don't use
> tabs :-) ).
>
>
> > @@ -129,7 +129,8 @@
> >              state = 1;
> >          }
> >
> > -        if ((state == 1) && !strcmp(str, "\n")) {
> > +	/* Str can be '\n' or CRLF */
> > +        if ((state == 1) && (!strcmp(str, "\n") || !strcmp(str, CRLF)))
{
> >              msg->header_end = off;
> >              msg->msg_start = off + 1;
> >             state = 2;
>
> If str==CRLF, then shouldn't msg_start=off+2 ?
> (And again, watch those tabs.)
>
>
> > @@ -91,6 +91,7 @@
> >
> >  static char *compute_md5(request_rec *r, pop_msg *msg)
> >  {
> > +    apr_status_t av;
> >      apr_mmap_t *mm = NULL;
> >      apr_finfo_t finfo;
> >     pop_user_rec *ur = (pop_user_rec *)ap_get_module_config(r->request_config,
>
> A very minor nit: by convention we usually call our apr_status_t's "rv"
> rather than "av".
>
>
> >      apr_md5_init(ur->ctx);
> > -    apr_md5_update(ur->ctx, (char*)mm->mm + msg->header_start, msg->msg_end
- msg->header_start);
> > +    apr_md5_update(ur->ctx, (char*)mm->mm + msg->header_start,
> > +		   msg->msg_end - msg->header_start);
> > +    apr_mmap_delete(mm);
>
> Again the note on mixing stylistic changes and code changes.  I agree that
> this is a good stylistic change, but let's clean up the style in a
> separate patch.  :)
>
> @@ -148,6 +154,20 @@
>              continue;
>          }
>          res = handle_func->func(r, buffer);
> > +
> > +	/* Fill request_rec for access log */
> > +	if (!strcmp(command, "pass"))
> > +	    r->the_request = apr_psprintf(r->pool, "POP3/%s xxxxxx", command);
> > +	else
> > +	    r->the_request = apr_psprintf(r->pool, "POP3/%s %s", command, buffer);
>
> Very minor point: If you already know command=="pass", you might as well
> just say "POP3/pass xxxxx" instead of using %s, command.  Also very minor
> point:  Please put then and else conditions inside {}'s, even if they are
> just one line (helps make the code future-proof).
>
> > +
> > +	ap_rflush(r);
> >         ap_rwrite(".\r\n", strlen(".\r\n"), r);
> > -        ap_rflush(r);
> > +	ap_rflush(r);
> >      }
> >      else {
> >          generate_scan_listing(r, ur->mbox, atoi(num), 1);
>
> I'm very unclear why this additional flush is necessary.  I believe you
> that it is, however it would seem to indicate a bug elsewhere in the httpd
> (like in the buffering code that handles ap_r*() calls).  I'll have a peek
> at that code and mention the problem on dev@httpd so that we can maybe
> find the real source of the problem instead of just working around it.
>
> Thanks again for your help!
>
> --Cliff
>

Re: [PATCH] Logging, Two dots, MMAP, etc

2003-09-27T20:46:52Z

On Sat, 27 Sep 2003, Sung Kim wrote:

> 1. Logging for each POP3 command and return status.
> 2. Two dots from LIST and TOP commands.
> 3. Error on apr_mmap_create. apr_mmap_create will return error for
> APR_BURRERED file.
> 4. Added file_close and apr_mmap_delete for cleanup.

Hey, thanks for the patch!  A few minorcomments inline:

> +++ pop_mbox.c	27 Sep 2003 06:50:40 -0000
> @@ -65,7 +65,7 @@
>  apr_status_t pop_parse_maildrop(request_rec *r, pop_mbox **mbox)
>  {
>      apr_status_t rv;
> -    char *str = apr_pcalloc(r->pool, POP_STRING_LENGTH);
> +    char str[POP_STRING_LENGTH];
>      apr_off_t off = 0;
>      int state = 0;
>      int messages = 1;

What's the purpose of this change?  Are you seeing a memory leak when
pcalloc is used?  We tend to prefer that against stack buffers whenever
possible.

> @@ -83,7 +83,7 @@
>
>      ap_run_translate_name(r);
>
> -    if ((rv = apr_file_open(&ur->fp, r->filename, APR_READ | APR_WRITE | APR_BUFFERED,
> +    if ((rv = apr_file_open(&ur->fp, r->filename, APR_READ | APR_WRITE,
>                         APR_OS_DEFAULT, ur->p)) != APR_SUCCESS) {
>          ap_fprintf(r->output_filters, bb,
>                     "-ERR, unable to open %s's maildrop.\n", ur->user);

Good catch!

> @@ -111,7 +111,7 @@
>
>          rv = apr_file_gets(str, bytes, ur->fp);
>          if (rv != APR_SUCCESS && rv != APR_EOF) {
> -            return rv;
> +	    return rv;
>          }
>
>          if ((!strncmp(str, "From ", strlen("From "))) || rv == APR_EOF) {

Please try to avoid stylistic changes in functional patches; it makes them
harder to review.  Also please refer to the coding styleguide at
http://httpd.apache.org/dev/styleguide.html (in particular, don't use
tabs :-) ).

> @@ -129,7 +129,8 @@
>              state = 1;
>          }
>
> -        if ((state == 1) && !strcmp(str, "\n")) {
> +	/* Str can be '\n' or CRLF */
> +        if ((state == 1) && (!strcmp(str, "\n") || !strcmp(str, CRLF))) {
>              msg->header_end = off;
>              msg->msg_start = off + 1;
>             state = 2;

If str==CRLF, then shouldn't msg_start=off+2 ?
(And again, watch those tabs.)

> @@ -91,6 +91,7 @@
>
>  static char *compute_md5(request_rec *r, pop_msg *msg)
>  {
> +    apr_status_t av;
>      apr_mmap_t *mm = NULL;
>      apr_finfo_t finfo;
>     pop_user_rec *ur = (pop_user_rec *)ap_get_module_config(r->request_config,

A very minor nit: by convention we usually call our apr_status_t's "rv"
rather than "av".

>      apr_md5_init(ur->ctx);
> -    apr_md5_update(ur->ctx, (char*)mm->mm + msg->header_start, msg->msg_end
- msg->header_start);
> +    apr_md5_update(ur->ctx, (char*)mm->mm + msg->header_start,
> +		   msg->msg_end - msg->header_start);
> +    apr_mmap_delete(mm);

Again the note on mixing stylistic changes and code changes.  I agree that
this is a good stylistic change, but let's clean up the style in a
separate patch.  :)

@@ -148,6 +154,20 @@
             continue;
         }
         res = handle_func->func(r, buffer);
> +
> +	/* Fill request_rec for access log */
> +	if (!strcmp(command, "pass"))
> +	    r->the_request = apr_psprintf(r->pool, "POP3/%s xxxxxx", command);
> +	else
> +	    r->the_request = apr_psprintf(r->pool, "POP3/%s %s", command, buffer);

Very minor point: If you already know command=="pass", you might as well
just say "POP3/pass xxxxx" instead of using %s, command.  Also very minor
point:  Please put then and else conditions inside {}'s, even if they are
just one line (helps make the code future-proof).

> +
> +	ap_rflush(r);
>         ap_rwrite(".\r\n", strlen(".\r\n"), r);
> -        ap_rflush(r);
> +	ap_rflush(r);
>      }
>      else {
>          generate_scan_listing(r, ur->mbox, atoi(num), 1);

I'm very unclear why this additional flush is necessary.  I believe you
that it is, however it would seem to indicate a bug elsewhere in the httpd
(like in the buffering code that handles ap_r*() calls).  I'll have a peek
at that code and mention the problem on dev@httpd so that we can maybe
find the real source of the problem instead of just working around it.

Thanks again for your help!

--Cliff

[PATCH] Logging, Two dots, MMAP, etc

2003-09-27T07:21:53Z

Dear Httpd-pop3 developers,

Recently, I found several problems in the module and this patch includes:

1. Logging for each POP3 command and return status.
2. Two dots from LIST and TOP commands.
3. Error on apr_mmap_create. apr_mmap_create will return error for APR_BURRERED file.
4. Added file_close and apr_mmap_delete for cleanup.

Have a nice day.

--
Sung Kim 
http://www.cse.ucsc.edu/~hunkim

 "Dreams become reality!"

[httpd-pop3] UIDL Segment fault

2003-09-27T03:05:15Z

Hello,

httpd_pop3 has problem with the UIDL command. It seems like that mm is
NULL after apr_mmap_create, even though the function returns APR_SUCESS.
Any idea?

- 
Sung Kim

(gdb) r -X
Starting program: /usr/local/apache2/bin/httpd -X
[New Thread 8192 (LWP 17911)]

Program received signal SIGSEGV, Segmentation fault.
[Switching to Thread 8192 (LWP 17911)]
0x080673b5 in compute_md5 (r=0x8173460, msg=0x81725e0) at
pop_protocol.c:109
109         apr_md5_update(ur->ctx, (char*)mm->mm + msg->header_start,
msg->msg_end - msg->header_start);
(gdb) p mm
$1 = (struct apr_mmap_t *) 0x0
(gdb) where
#0  0x080673b5 in compute_md5 (r=0x8173460, msg=0x81725e0) at
pop_protocol.c:109
#1  0x0806767f in generate_unique_listing (r=0x8173460, mbox=0x8174018,
num=1, with_ok=0) at pop_protocol.c:241
#2  0x08067e7e in ap_handle_uidl (r=0x8173460, buffer=0x8173cfe "") at
pop_protocol.c:566
#3  0x080674b6 in process_pop_connection_internal (r=0x8173460,
bb=0x81714f0) at pop_protocol.c:154
#4  0x08066f64 in process_pop_connection (c=0x81714f0) at pop_core.c:148
#5  0x08096132 in ap_run_process_connection (c=0x8169990) at
connection.c:85
#6  0x0808beef in child_main (child_num_arg=0) at prefork.c:694
#7  0x0808c09a in make_child (s=0x810f148, slot=0) at prefork.c:734
#8  0x0808c0f3 in startup_children (number_to_start=5) at prefork.c:806
#9  0x0808c7e5 in ap_mpm_run (_pconf=0x808b7d8, plog=0x8109bc8,
s=0x810f148) at prefork.c:1022
#10 0x080914fe in main (argc=2, argv=0xbffffad4) at main.c:660
#11 0x420158d4 in __libc_start_main () from /lib/i686/libc.so.6

Tow dots problem?

2003-09-27T00:45:05Z

Hello,

I tried to use httpd_pop and my pop3 client just hang. I figured the 
list command returns two dots:

list
+OK 4 messages (1608 octets)
1 401
2 410
3 394
4 399
..

Also top commands returns two dots as well:
top 1 2
+OK
 From hunkim@ocean.cse.ucsc.edu Fri Sep 26 15:19:50 2003
Received: (from hunkim@localhost)
        by ocean.cse.ucsc.edu (8.11.6/8.11.4) id h8QNGdM02092
        for hunkim@catacomb.mine.nu; Fri, 26 Sep 2003 16:16:39 -0700
Date: Fri, 26 Sep 2003 16:16:39 -0700
From: Sung Kim 
Message-Id: <200309262316.h8QNGdM02092@ocean.cse.ucsc.edu>
To: hunkim@catacomb.mine.nu
Subject: test
 
 
wefewf
..

But adding ap_rflush(r) after rprint functions, the problem was solved.
Do you have any idea on that? I guess pop_crlf_esacpe_filter does 
something, but not sure.

- Sung Kim

[PATCH] for access_log

2003-09-25T07:53:11Z

Hello,

This patch enables httpd-pop3 to keeps the access_log.

The log example is below:
127.0.0.1 - hunkim [25/Sep/2003:00:43:11 -0700] "POP3/user hunkim" - -
127.0.0.1 - hunkim [25/Sep/2003:00:43:11 -0700] "POP3/pass xxxxxx" - -
127.0.0.1 - hunkim [25/Sep/2003:00:43:14 -0700] "POP3/stat " - -
127.0.0.1 - hunkim [25/Sep/2003:00:43:14 -0700] "POP3/quit " 1 -

Index: pop_protocol.c
===================================================================
RCS file: /home/cvspublic/httpd-pop3/pop_protocol.c,v
retrieving revision 1.10
diff -u -r1.10 pop_protocol.c
--- pop_protocol.c      22 Jan 2003 04:55:04 -0000      1.10
+++ pop_protocol.c      25 Sep 2003 07:45:59 -0000
@@ -148,6 +148,20 @@
             continue;
         }
         res = handle_func->func(r, buffer);
+
+       /* Fill request_rec for access log */
+       if (!strcmp(command, "pass"))
+           r->the_request = apr_psprintf(r->pool, "POP3/%s xxxxxx", command);
+       else
+           r->the_request = apr_psprintf(r->pool, "POP3/%s %s", command, buffer);
+
+       r->request_time = apr_time_now();
+       r->status = res;
+       ap_run_map_to_storage(r);
+
+       /* Write access log */
+       ap_run_log_transaction(r);
+
         if (res == POP_QUIT) {
             break;
         }
@@ -355,6 +369,7 @@
         rewrite_mbox(r);
         msg_num_and_size(ur->mbox, &num, &size);
         apr_file_unlock(ur->fp);
+        apr_file_close(ur->fp);
         ap_rprintf(r, "+OK %s POP3 server signing off ", ap_get_server_name(r));
         if (num == 0) {
             ap_rputs("(maildrop empty)\r\n", r);


--
Sung Kim 
http://www.cse.ucsc.edu/~hunkim

 "Dreams become reality!"

RE: Is anyone working on mod_smtp?

2003-09-24T20:39:33Z

At 12:19 PM 9/24/2003, Sung Kim wrote:
>Hi William, 
>
>Glad to hear mod_pop3 is going to be a pert of Apache 2.1.

It won't be folded in, but by keeping it in sync we will keep it out there
as an example for other protocol module authors :-)

>Dose mod_pop3 keeps the access log? Where can I send a patch for
>mod_pop3, if I have any?

I believe yes - access is logged, but I could be mistaken.

Patches should come right here to this list.

>I went to http://www.apache.org/foundation/mailinglists.html, and
>couldn't find the mail archive for pop3. Also I was wondering there is a
>work for mod_smtp. I'm pretty much interest in either joining the
>development team or developing mod_smtp.

It sure seems like this would be a good place for the initial discussion,
at least.  I will search for mbox files for you.

Bill

RE: Is anyone working on mod_smtp?

2003-09-24T18:02:44Z

At 12:19 PM 9/24/2003, Sung Kim wrote:
>Hi William, 
>
>Glad to hear mod_pop3 is going to be a pert of Apache 2.1.

It won't be folded in, but by keeping it in sync we will keep it out there
as an example for other protocol module authors :-)

>Dose mod_pop3 keeps the access log? Where can I send a patch for
>mod_pop3, if I have any?

I believe yes - access is logged, but I could be mistaken.

Patches should come right here to this list.

>I went to http://www.apache.org/foundation/mailinglists.html, and
>couldn't find the mail archive for pop3. Also I was wondering there is a
>work for mod_smtp. I'm pretty much interest in either joining the
>development team or developing mod_smtp.

It sure seems like this would be a good place for the initial discussion,
at least.  I will search for mbox files for you.

Bill

RE: Is anyone working on mod_smtp?

2003-09-24T17:19:58Z

Hi William, 

Glad to hear mod_pop3 is going to be a pert of Apache 2.1.
Dose mod_pop3 keeps the access log? Where can I send a patch for
mod_pop3, if I have any?

I went to http://www.apache.org/foundation/mailinglists.html, and
couldn't find the mail archive for pop3. Also I was wondering there is a
work for mod_smtp. I'm pretty much interest in either joining the
development team or developing mod_smtp.

Have a nice day.

- Sung
-----Original Message-----
From: William A. Rowe, Jr. [mailto:wrowe@rowe-clan.net] 
Sent: Wednesday, September 24, 2003 9:47 AM
To: Sung Kim
Cc: pop3-dev@httpd.apache.org
Subject: Re: Is anyone working on mod_smtp? 

At 07:07 PM 9/23/2003, Sung Kim wrote:

>Hello,
>
>I installed httpd-pop3 today, and its very neat. The problem is getting
a mail file from MTA. Also there are some security issues, and the
username of system and Apache authentication should be synchronized. 
>
>If there is a mod_smtp and get emails for Apache authenticated user, it
will work very well with httpd-pop3.

Glad to find ongoing interest in this module :-)

>Is anyone working on mod_smtp or something?

mod_pop3 was truly developed as an example of protocol extensions
within Apache.  The module will likely be updated for the httpd-2.1 dev
versions (httpd-2.0's API was frozen - and the next public releases
after the 2.0 series will be the 2.2 series.)

There are several of us here with alot of committments, but I'm looking
forward to making sure this module builds under 2.1.  As with all Apache
projects - patches are *always* welcomed.

>PS How can I access
pop3-dev@httpd.apache.org mail
archive?

See http://www.apache.org/foundation/mailinglists.html for some more
info
and let us know if that doesn't help you out.

Bill

Re: Is anyone working on mod_smtp?

2003-09-24T16:46:34Z

At 07:07 PM 9/23/2003, Sung Kim wrote:

>Hello,
>
>I installed httpd-pop3 today, and its very neat. The problem is getting a mail file from
MTA. Also there are some security issues, and the username of system and Apache authentication
should be synchronized. 
>
>If there is a mod_smtp and get emails for Apache authenticated user, it will work very
well with httpd-pop3.

Glad to find ongoing interest in this module :-)

>Is anyone working on mod_smtp or something?

mod_pop3 was truly developed as an example of protocol extensions
within Apache.  The module will likely be updated for the httpd-2.1 dev
versions (httpd-2.0's API was frozen - and the next public releases
after the 2.0 series will be the 2.2 series.)

There are several of us here with alot of committments, but I'm looking
forward to making sure this module builds under 2.1.  As with all Apache
projects - patches are *always* welcomed.

>PS How can I access pop3-dev@httpd.apache.org
mail archive?

See http://www.apache.org/foundation/mailinglists.html for some more info
and let us know if that doesn't help you out.

Bill

Is anyone working on mod_smtp?

2003-09-24T00:07:10Z

Hello,
 
I installed httpd-pop3 today, and it's very neat. The problem is getting
a mail file from MTA. Also there are some security issues, and the
username of system and Apache authentication should be synchronized. 
If there is a mod_smtp and get emails for Apache authenticated user, it
will work very well with httpd-pop3.
 
Is anyone working on mod_smtp or something?
 
PS - How can I access pop3-dev@httpd.apache.org mail archive?
--
Sung Kim, UCSC

RE: is this module integrated with handlers at all

2003-07-17T15:33:16Z


i found out that modpython actually already has connection handlers for
making protocols

it wasnt 100% functional but a small patch fixed it

the url to the echo sample is here.. its pretty easy to start, but
there's
2 small patches that may or may not have been integrated into 3.0.3 for
it
to log errors properly right now

http://www.modpython.org/live/current/doc-html/pyapi-conn.html


> -----Original Message-----
> From: William A. Rowe, Jr. [mailto:wrowe@rowe-clan.net] 
> Sent: Tuesday, July 15, 2003 4:10 PM
> To: k d
> Cc: pop3-dev@httpd.apache.org
> Subject: Re: is this module integrated with handlers at all
> 
> 
> It is sort of in it's infancy... and certainly could use all 
> the help we can get
> in really rounding it out.
> 
> As a matter of fact, I'm looking at a 'vanilla' telnet 
> handler (filter and
> bucket type) that could hand the lightweight requirements of other
> protocols such as smtp, ftp, etc.  This would simply comply with the
> telnet spec (FF over the wire as FFFF, create a custom bucket for
> telnet metadata such as IP, SYNC etc.)
> 
> As far as handler integration, I believe this module doesn't 
> at the moment.
> I wish it would and might help if I free up some time next month.
> 
> Bill
> 
> At 02:40 PM 7/15/2003, k d wrote:
> >i don't really see this as being a transparent protocol 
> implementation..
> >
> >of course that's probably not what it was designed for, but 
> that's the functionality that i think makes the most sense 
> for any 'protocol modules'..
> >
> >as a stand-alone module that has to be fully developed in C, 
> i don't see how this would be too terribly useful..
> >
> >if it were however transparently accessable through handles, 
> then it could be integrated with existing apache2 code for 
> other services..
> >
> >not that i have any clue atm how that architecture would 
> work out, but i am looking for that kind of functionality in 
> a telnet protocol module..
> >
> >_________________________________________________________________
> >The new MSN 8: advanced junk mail protection and 2 months FREE*  
> >http://join.msn.com/?page=features/junkmail
> >
> 
> 
>

Re: is this module integrated with handlers at all

2003-07-15T20:50:17Z

hmm im not familiar enough with any of the parts of apache2 at the moment to 
really comment

just started looking at the protocol modules today and noticed that they 
seem to be standalone as opposed to integrated...

what i would find useful (but may not be possible) is being able to use 
mod_python as the driver for not only a www interface but also telnet and 
pop3/imap.. where i am drawing blanks right now (mostly because i haven't 
spent any time trying to think about it design wise) is how the mod_python 
handlers integrate with the protocol modules to obtain access to extended 
services within the same framework

i am assuming right now with an adequate low level transparent protocol 
interface that would support the right line feeds and communication 
interface that it would be easy to implement handlers in python for the 
various protocols.. i guess telnet is the base for most of them given either 
char mode or line mode or something.. so with some python code on top of 
that itd be easy to have pop3 or something like a telnet gateway where www 
data can be displayed in telnet fashion, either raw or interactive.. like a 
telnet bbs for instance.. but same user accounts and base code can implement 
the bbs, the mail, and say forums on the www.. with persistent db connection 
pooling and everything else already available in apache2

i'm mostly focusing on the telnet stuff right now because my pop3 client a 
long time ago hit a brick wall with outlook's mime parsing..  but i'm trying 
to get a grasp on the protocol module implementation to see if its robust 
enough to even bother pursuing

>From: "William A. Rowe, Jr." 
>To: "k d" 
>CC: pop3-dev@httpd.apache.org
>Subject: Re: is this module integrated with handlers at all
>Date: Tue, 15 Jul 2003 15:10:01 -0500
>
>It is sort of in it's infancy... and certainly could use all the help we 
>can get
>in really rounding it out.
>
>As a matter of fact, I'm looking at a 'vanilla' telnet handler (filter and
>bucket type) that could hand the lightweight requirements of other
>protocols such as smtp, ftp, etc.  This would simply comply with the
>telnet spec (FF over the wire as FFFF, create a custom bucket for
>telnet metadata such as IP, SYNC etc.)
>
>As far as handler integration, I believe this module doesn't at the moment.
>I wish it would and might help if I free up some time next month.
>
>Bill
>
>At 02:40 PM 7/15/2003, k d wrote:
> >i don't really see this as being a transparent protocol implementation..
> >
> >of course that's probably not what it was designed for, but that's the 
>functionality that i think makes the most sense for any 'protocol 
>modules'..
> >
> >as a stand-alone module that has to be fully developed in C, i don't see 
>how this would be too terribly useful..
> >
> >if it were however transparently accessable through handles, then it 
>could be integrated with existing apache2 code for other services..
> >
> >not that i have any clue atm how that architecture would work out, but i 
>am looking for that kind of functionality in a telnet protocol module..
> >
> >_________________________________________________________________
> >The new MSN 8: advanced junk mail protection and 2 months FREE*
> >http://join.msn.com/?page=features/junkmail
> >
>
>

_________________________________________________________________
The new MSN 8: advanced junk mail protection and 2 months FREE*  
http://join.msn.com/?page=features/junkmail

Re: is this module integrated with handlers at all

2003-07-15T20:10:01Z

It is sort of in it's infancy... and certainly could use all the help we can get
in really rounding it out.

As a matter of fact, I'm looking at a 'vanilla' telnet handler (filter and
bucket type) that could hand the lightweight requirements of other
protocols such as smtp, ftp, etc.  This would simply comply with the
telnet spec (FF over the wire as FFFF, create a custom bucket for
telnet metadata such as IP, SYNC etc.)

As far as handler integration, I believe this module doesn't at the moment.
I wish it would and might help if I free up some time next month.

Bill

At 02:40 PM 7/15/2003, k d wrote:
>i don't really see this as being a transparent protocol implementation..
>
>of course that's probably not what it was designed for, but that's the functionality that
i think makes the most sense for any 'protocol modules'..
>
>as a stand-alone module that has to be fully developed in C, i don't see how this would
be too terribly useful..
>
>if it were however transparently accessable through handles, then it could be integrated
with existing apache2 code for other services..
>
>not that i have any clue atm how that architecture would work out, but i am looking for
that kind of functionality in a telnet protocol module..
>
>_________________________________________________________________
>The new MSN 8: advanced junk mail protection and 2 months FREE*  
>http://join.msn.com/?page=features/junkmail
>

is this module integrated with handlers at all

2003-07-15T19:40:37Z

i don't really see this as being a transparent protocol implementation..

of course that's probably not what it was designed for, but that's the 
functionality that i think makes the most sense for any 'protocol modules'..

as a stand-alone module that has to be fully developed in C, i don't see how 
this would be too terribly useful..

if it were however transparently accessable through handles, then it could 
be integrated with existing apache2 code for other services..

not that i have any clue atm how that architecture would work out, but i am 
looking for that kind of functionality in a telnet protocol module..

_________________________________________________________________
The new MSN 8: advanced junk mail protection and 2 months FREE*  
http://join.msn.com/?page=features/junkmail

RE: [PATCH] static compile change

2002-07-17T03:40:29Z

Fixing this another way.  Basically, my instructions were wrong.  :-(

Ryan

----------------------------------------------
Ryan Bloom                  rbb@covalent.net
645 Howard St.              rbb@apache.org
San Francisco, CA 

> -----Original Message-----
> From: Thomas Eibner [mailto:thomas@stderr.net]
> Sent: Tuesday, July 16, 2002 3:52 PM
> To: pop3-dev@httpd.apache.org
> Subject: [PATCH] static compile change
> 
> Without the change configure tries to create modules/httpd-pop3 and
> breaks when you `make'
> 
> Index: config.m4
> ===================================================================
> RCS file: /home/cvspublic/httpd-pop3/config.m4,v
> retrieving revision 1.2
> diff -r1.2 config.m4
> 3c3
> < APACHE_MODPATH_INIT(httpd-pop3)
> ---
> > APACHE_MODPATH_INIT(pop)
> 
> --
>   Thomas Eibner  DnsZone

>   mod_pointer 

>   !(C)

>           Putting the HEST in .COM

[PATCH] static compile change

2002-07-16T22:52:15Z

Without the change configure tries to create modules/httpd-pop3 and
breaks when you `make'

Index: config.m4
===================================================================
RCS file: /home/cvspublic/httpd-pop3/config.m4,v
retrieving revision 1.2
diff -r1.2 config.m4
3c3
< APACHE_MODPATH_INIT(httpd-pop3)
---
> APACHE_MODPATH_INIT(pop)

-- 
  Thomas Eibner  DnsZone 
  mod_pointer  
  !(C)                  
          Putting the HEST in .COM

RE: 2 questions...

2001-07-08T21:57:51Z

On Sun, 8 Jul 2001, Sean R. Bright wrote:

> My understanding is that its not pop_process_connection's responsibility to
> decline HTTP sessions.  You need stricter Listen/VirtualHost directives.
>
> This is my setup:
> ===========================><8===========================
> ###########################################################
> # Listen
> ###########################################################
> Listen 24.6.145.136:80
> Listen 192.168.0.24:80
>
> Listen 24.6.145.136:110
> Listen 192.168.0.24:110
>
> ###########################################################
> # VirtualHosts
> ###########################################################
> 
>     ServerAdmin blah@host.com
>     DocumentRoot /usr/local/etc/httpd-2.0/htdocs
>     ServerName myhost.com
>     ErrorLog logs/error_log
>     CustomLog logs/access_log common
> 
>
> 
>     POP3Protocol        On
>     POP3MailDrops       /var/spool/mail
> 
> ===========================8><===========================
>
> Works fine for me.  Requests coming in on 110 are handled by mod_pop and the
> rest are handled by mod_http.

That's about right.  I should warn you, if your mail has a '.' as the
first character of a line, the POP3 server will not escape it properly.
This is likely to break your client app.

I have the solution, I just need to test and commit it.

Ryan

_____________________________________________________________________________
Ryan Bloom                        	rbb@apache.org
Covalent Technologies			rbb@covalent.net
-----------------------------------------------------------------------------

RE: 2 questions...

2001-07-08T21:05:48Z

My understanding is that its not pop_process_connection's responsibility to
decline HTTP sessions.  You need stricter Listen/VirtualHost directives.

This is my setup:
===========================><8===========================
###########################################################
# Listen
###########################################################
Listen 24.6.145.136:80
Listen 192.168.0.24:80

Listen 24.6.145.136:110
Listen 192.168.0.24:110

###########################################################
# VirtualHosts
###########################################################

    ServerAdmin blah@host.com
    DocumentRoot /usr/local/etc/httpd-2.0/htdocs
    ServerName myhost.com
    ErrorLog logs/error_log
    CustomLog logs/access_log common



    POP3Protocol        On
    POP3MailDrops       /var/spool/mail

===========================8><===========================

Works fine for me.  Requests coming in on 110 are handled by mod_pop and the
rest are handled by mod_http.

Sean

> -----Original Message-----
> From: Sapan.Bhatia@nokia.com [mailto:Sapan.Bhatia@nokia.com]
> Sent: Sunday, July 08, 2001 3:51 PM
> To: rbb@covalent.net
> Cc: pop3-dev@httpd.apache.org
> Subject: RE: 2 questions...
>
>
> //The HTTP process_connection phase is registered as
> HOOK_REALLY_LAST, so
> //it
> //is always the last process_connection phase run.  If you create a
> //Virtual
> //host, and setup POP to run on it, the
> pop_process_connection phase will
> //run first, and it will handle the connection.  Let me know if that
> //doesn't
> //explain it.  :-)
>
> Fine.
>
> But then how does pop_process_connection decline HTTP sessions??
>
> This is something I'd struggled with for many hours, and
> finally had to
> resort to declining on the basis of base_server->port... so
> I'll be annoyed
> and relieved if there's a better solution...
>
> -Sapan
>

RE: 2 questions...

2001-07-08T19:51:17Z

//The HTTP process_connection phase is registered as HOOK_REALLY_LAST, so 
//it 
//is always the last process_connection phase run.  If you create a 
//Virtual 
//host, and setup POP to run on it, the pop_process_connection phase will 
//run first, and it will handle the connection.  Let me know if that 
//doesn't 
//explain it.  :-) 

Fine. 

But then how does pop_process_connection decline HTTP sessions?? 

This is something I'd struggled with for many hours, and finally had to
resort to declining on the basis of base_server->port... so I'll be annoyed
and relieved if there's a better solution...

-Sapan

POP3 code committed.

2001-07-08T17:14:17Z


In case anybody cares, I have committed the POP3 module.  I don't have
docs that tell you how to turn it on yet, but the code is there.  It is
relatively easy to follow if you understand the HTTPD code.  I'll continue
to work on this as I have time, but the more people who get involved, the
better.

Ryan


_____________________________________________________________________________
Ryan Bloom                        	rbb@apache.org
Covalent Technologies			rbb@covalent.net
-----------------------------------------------------------------------------

2 questions...

2001-07-08T14:21:27Z

Hello,

I tried compiling mod_pop with Apache 2_0_16, and it was giving a few
compilation errors - eg. extra int& argument to get_brigade and a call to a
nonexistant apr_file_trunc...

Also - more importantly, considering that ap_process_http_connection will be
registered prior to ap_process_pop_connection, shouldn't
ap_process_http_connection somehow decline a POP3 session...  eg. on the
basis of connection->base_server->port... 

Or how would it work otherwise...?

-Sapan