[AOLSERVER] Question about extending AOLserver in C
So I'm studying set.c and tclset.c, because the ns_set is probably the extension I use the most and I figure it's Done Right -- but I can't for the life of me see how the connection-local interpreter knows to call Ns_TclFreeSet. Where is the proper cleanup function assigned to the interpreter? I'm so confused Michael
Re: [AOLSERVER] AOLSERVER Digest - 22 Aug 2001 to 23 Aug 2001 (#2001-215)
In a message dated 8/24/01 12:01:16 AM Eastern Daylight Time, [EMAIL PROTECTED] writes: Fw: AOLserver 3.0 AOLserver 3.0 vulnerability The 3.0 vulnerability refers to a buffer overflow error that can be the focus of a DOS attack. The error occurs while parsing the authorization header; the buffer is created on the stack in ParseAuth() and is potentially overflowed in Ns_HtuuDecode(). This was reported and patched (thanks to Adam Zell!) for the 3.4pre1 release. Jay Ridgeway AOL Local Technologies
[AOLSERVER] How to know the language from a user?
Hi all, Some knows how to know the language from a user, when this user go to your main page http://www.?.com for first time?. Without any cookie or user_id. I guess there are to possibilities, from the user_IP and then nslookup, or from the browser_language. I guess is better the second option. Does one tcl function exist to execute from a index.tcl file to get this kind of information? I'm running aolserver3.1+ad8+nsvhr, with ACS 3.4 on suse Linux. Thank you in advance and best regards, Antoni Vila
[AOLSERVER] SOAP
Hello. Some time ago, someone (yes, I have a bd memory :) mentioned about porting tclsoap to AOLserver. Where can I read about it, join the project or something? I don't want to double the code/efforts, but I'd like to use client-server architecture for managing users/articles/that stuff on the server. It's a bit of a mess to do it via the browser, but a lot easier with client-server. Thanks for your help :) -- WK
[AOLSERVER] Logging: reverse dns lookups and file access permissions
I have a patch at http://www.theashergroup. com/download, ns_log_modes, that modifies nsd/log.c to examine a new config.tcl parameter ns_section ns/parameters ns_param serverlogmode 0664 nsd/log.c will open the server log and then change its mode to serverlogmode. Serverlogmode defaults to 0644. Also, the access log has a similar change: ns_section ns/server/yourserver/module/nslog ns_param accessmode 0664 ns_param iplookups true nslog/nslog.c will open the access log and then change its mode to accessmode. Accessmode also defaults to 0644. I find that opening logs to different modes is useful for ISPs or corporate installations of AOLserver, in which different users, at different times, may need to start/stop AOLserver, and/or delete the log files. I also added a new parameter, iplookups that defaults to off. If set to on, then the access log will perform Reverse DNS Lookups for you. The usual caveat applies, that this can slow down overall AOLserver performance. = Jerry Asher [EMAIL PROTECTED] 1678 Shattuck Avenue Suite 161Tel: (510) 549-2980 Berkeley, CA 94709Fax: (877) 311-8688
[AOLSERVER] member site charging
Those of you who were interested in my experiences with setting up ccBill to handle credit card charging for a member based site with Aolserver please email be offlist at [EMAIL PROTECTED] as I am in the process of writing up my description of what I did, and collecting all the resources in one location.. this way I can give those interested a full description of what I did and how I got it to work... Enjoy... Patrick
[AOLSERVER] patch for html 4.0, use semicolons and not ampersands in uris
I have created a patch, semicolons.in.uris for AOLserver at http://www.theashergroup.com/download (it's also available from sourceforge) to support HTML4.0 specification Appendix B.2.2, regarding special characters in URL attribute values, specifically using a semicolon, ';', in a URI instead of an ampersand, ''. With this patch, AOLserver will take a URI that uses semicolons and not ampersands and correctly parse the URI into form variables and their values. http://www.w3.org/TR/REC-html40/appendix/notes.html#h-B.2.2 quote B.2.2 Ampersands in URI attribute values The URI that is constructed when a form is submitted may be used as an anchor-style link (e.g., the href attribute for the A element). Unfortunately, the use of the character to separate form fields interacts with its use in SGML attribute values to delimit character entity references. For example, to use the URI http://host/?x=1y=2; as a linking URI, it must be written A href=http://host/?x=1#38;y=2; or A href=http://host/?x=1amp;y=2;. We recommend that HTTP server implementors, and in particular, CGI implementors support the use of ; in place of to save authors the trouble of escaping characters in this manner. /quote This patch adds a config.tcl parameter ns_section ns/server/yourserver ns_param semicolonqueryseparator yes The default value is no, or not examining URI's for semicolons. nsd/nsconf.c - read ns_param semicolonseparator for ns_querytoset nsd/nsconf.h - define CONN_SEMICOLON_SEP_BOOL for ns_querytoset nsd/nsd.h - added semicolonseparator to struct conn in struct _nsconf for ns_querytoset nsd/conn.c - NS_QueryToSet - modified to examine semicolons inside of a request = Jerry Asher [EMAIL PROTECTED] 1678 Shattuck Avenue Suite 161Tel: (510) 549-2980 Berkeley, CA 94709Fax: (877) 311-8688
[AOLSERVER] patch to dump arbitary hex/ascii to server log
I have a patch, ns_logdump, at http://www.theashergroup.com/download/, that adds a Tcl and C API to AOLserver to dump arbitrary hex/ascii strings into the log file. It's also available at sourceforge. Dump a string in hex and ascii to the log. Available from C or Tcl. Useful in debugging new protocols Add a C API and Tcl API for dumping HEX to the log. void Ns_LogDumpString(Ns_LogSeverity severity, char *s, int len) and ns_logdump severity string Severity is a standard ns_log severity constant, Notice, Warning, Error, Debug, or Dev. The output in the log file looks like: : 4164 6420 6120 4320 4150 4920 616e 6420 Add a C API and 0010: 5463 6c20 4150 4920 666f 7220 6475 6d70 Tcl API for dump 0020: 696e 6720 4845 5820 746f 2074 6865 206c ing HEX to the l 0030: 6f67 2e0a 0a76 6f69 6420 4e73 5f4c 6f67 og...void Ns_Log 0040: 4475 6d70 5374 7269 6e67 284e 735f 4c6f DumpString(Ns_Lo The length of a line is controlled by a tcl config parameter, dumpbytesperline, that defaults to 32. ns_section ns/parameters ns_paramdumpbytesperline 32 This patch modifies: nsd/log.c, nsd/nsconf.c, nsd/nsconf.h, nsd/nsd.h, nsd/tclcmds.c. I have found this useful in hooking the ACS up to new services (Tellme, Voxeo, SOAP), or when trying to determine how to ns_httpget binaries. = Jerry Asher [EMAIL PROTECTED] 1678 Shattuck Avenue Suite 161Tel: (510) 549-2980 Berkeley, CA 94709Fax: (877) 311-8688
[AOLSERVER] Question about form handling on POSTS
Hello,
I recently noticed that ns_conn form will return the query args from a
POST or the URL args on a GET, but in the case of a POST, will not return
both.
For example, if my action=test_form.adp?arg1=val1 arg1 will not be present
in [ns_conn form].
I have made a page which illustrates this and describes it in further
depth. Could someone take a look and see what they think? Does the HTTP
spec have anything to say about this? (i.e. is it not kosher to put query
args in a form's action?)
I've attached the .adp if you'd like to look it over.
Or, for minimal fuss, you can check it out at:
http://www.rustybrooks.org/test/test_form.adp
to view the ADP source, look at
http://www.rustybrooks.org/source/test/test_form.adp
--
Rusty Brooks : http://www.rustybrooks.org/
Spewing wisdom from every orifice
--
This ADP page shows a possible deficit in AOLServer's form handling. Most notably,
ns_conn form returns the query args on the URL line (in the case of a GET) iOR/i
the query args that were posted, in the case of a POST. To me, it seems like in the
case of a POST, you'd want to gather query args from both places, in cases like this
one,
where you are posting to a URL while passing query args. This is especially valuable
in
cases like this one, where the user might pass in a form initially via the URL, and
wish
for it to keep propagating when he presses submit. In this case, I think both arg1
and arg2
should have values when posted. It's not entirely clear what should happen when a
variable exists
in both the posted values and the command line, but this should probably not be
treated any
differently than when two variables of the same name are posted (as in the case of
multiple-select
boxes)
pPlease forward any comments to a
href=mailto:[EMAIL PROTECTED][EMAIL PROTECTED]/a or leave a
comment at the bottom of this page.
p
Values returned by ns_conn form
pre
%
set r [ns_conn form]
for { set a 0 } { $a [ns_set size $r] } { incr a } {
ns_puts [ns_set key $r $a] == [ns_set value $r $a]br
}
%
/pre
form method=post action=test_form.adp?arg1=val1
Value for arg1: input name=arg2 type=text value=val2 size=10
p
input value=Go Here to test POST type=submit
p
a href=test_form.adp?arg1=val1Go here to test GET/a
/form
[AOLSERVER] TEST --- IGNORE
Test
[AOLSERVER] It's working AOLSERVER-LIST ?
...because I don't receive any message since Ago.23. Ariel.
[AOLSERVER] [ aolserver-Bugs-455156 ] segfault in nsperm
Bugs item #455156, was opened at 2001-08-24 17:26 You can respond by visiting: http://sourceforge.net/tracker/?func=detailatid=103152aid=455156group_id=3152 Category: Modules: nsperm (Permissions) Group: aolserver3_4 Status: Open Resolution: None Priority: 5 Submitted By: Jeff Rogers (dvrsn) Assigned to: Nobody/Anonymous (nobody) Summary: segfault in nsperm Initial Comment: If the configuration parameter SkipLocks is on (the default) then calling ns_perm checkpass user pass with a user that exists (regardless of whether the password is correct or not) will segfault the server. The reason is quite simple: the mutex in the users TclHashTable is not initialized if skiplocks is set, but UserAuthProc does not check if skiplocks is set before attempting to lock the mutex. Checking skiplocks before attempting to manipulate the mutex (as its done nearly everywhere else in the code) fixes the segfault. SetPassCmd is similarly broken, but it should probably not be allowed at all if skiplocks is set. Also, Ns_MutexLock and Ns_LockMutex are both used (similarly for Unlock); it seems that the first should be preferres as LockMutex is from the 2.x compatability code. These reasons are likely partly behind the complete rewrite of nsperm for 4.x, but that has not yet been released and people running 3.4 or earlier may want to know about this problem. -- You can respond by visiting: http://sourceforge.net/tracker/?func=detailatid=103152aid=455156group_id=3152
[AOLSERVER] [ aolserver-Bugs-455657 ] ns_db verbose works on handle not pool
Bugs item #455657, was opened at 2001-08-26 20:38 You can respond by visiting: http://sourceforge.net/tracker/?func=detailatid=103152aid=455657group_id=3152 Category: Architecture: Server (nsd) Group: aolserver3_3 Status: Open Resolution: None Priority: 5 Submitted By: Jerry Asher (jerryasher) Assigned to: Nobody/Anonymous (nobody) Summary: ns_db verbose works on handle not pool Initial Comment: The ns_db verbose command operates on a db handle. It doesn't operate on the db pool as the documentation describes. It would be good to have a command that did operate on a db pool in its entirety. -- You can respond by visiting: http://sourceforge.net/tracker/?func=detailatid=103152aid=455657group_id=3152
[AOLSERVER] [ aolserver-Bugs-455777 ] ns_server urlstats is not documented
Bugs item #455777, was opened at 2001-08-27 07:34 You can respond by visiting: http://sourceforge.net/tracker/?func=detailatid=103152aid=455777group_id=3152 Category: Other: Documentation Group: aolserver3_4 Status: Open Resolution: None Priority: 5 Submitted By: Jerry Asher (jerryasher) Assigned to: Nobody/Anonymous (nobody) Summary: ns_server urlstats is not documented Initial Comment: http://www.aolserver.com/docs/tcldev/tapi- 119.htm#193657 doesn't mention ns_server urlstats. -- You can respond by visiting: http://sourceforge.net/tracker/?func=detailatid=103152aid=455777group_id=3152
[AOLSERVER] [ aolserver-Bugs-449744 ] conn driver freeproc never called
Bugs item #449744, was opened at 2001-08-10 02:02
You can respond by visiting:
http://sourceforge.net/tracker/?func=detailatid=103152aid=449744group_id=3152
Category: Architecture: Server (nsd)
Group: None
Status: Open
Resolution: None
Priority: 7
Submitted By: Jerry Asher (jerryasher)
Assigned to: Nobody/Anonymous (nobody)
Summary: conn driver freeproc never called
Initial Comment:
Communications drivers can have an Ns_ConnFreeProc.
Although it is not quite clear what this procedure
does, the legacy nsunix code choose to use it to free
various buffers, specifically, the conn input buffer.
But no one ever calls these freeing procedures.
My suggestion: define a routine similar to
conn.c/Ns_ConnClose (called Ns_ConnFree...) that calls
connPtr-drvPtr-freeProc (if it's not null).
Where to call it though? I am not certain.
I chose to call it in in serv.c/ConnRun, as the second
to last statement, right after NsRunCleanups and right
before the dstring is freed.
Is this the right place?
--
Comment By: Jerry Asher (jerryasher)
Date: 2001-08-27 08:20
Message:
Logged In: YES
user_id=20647
This patch contains fixes for four bugs:
1. Memory leak by any communication driver that implements
a driver
free proc: the bug is that this freeproc is never
called within
AOLserver.
Create conn.c/Ns_ConnFree to provide a connection a way
to call
the driver free proc (supported in registration of a
driver, but
never called within AOLserver.)
Call Ns_ConnFree in serv.c/ConnRun at end of
connection. Also
call Ns_ConnFree within drv.c whenever there is a
problem queueing
connection.
Declare Ns_ConnFree within ns.h
2. in nsd/drv.c/RunDriver, there are three bugs present in
which well
defined functions are called with the wrong
structures. (Or
called before the structures have been initialized.)
Call it the
curse of casting, right? No compiler warnings as the
*wrong*
structures are correctly cast into the right types.
To see this, the best way is to compare how similar
routines are
called or defined by the nssock driver.
Claim 1: In the excerpt
dPtr = arg;
dData = dPtr-drvData;
if (dPtr-locationProc != NULL) {
loc = (*dPtr-locationProc)(dData);
. . .
while ((status = ((*dPtr-acceptProc)(dData, cData)))
== NS_OK) {
A locationProc is supposed to take a connection structure,
if so, then
locationProc is being called before the accept, and so the
connection
data (presumably dData) has not yet been initialized.
(I'll make the
claim that dData is not necessarily the connection
structure that
locationproc needs, but regardless of whether it is or
isn't, it's not
fair to ask locationproc to crawl down the connection
structure before
the connection has been accepted.)
Here is how ns.h defines the locationProc:
typedef char *(Ns_ConnLocationProc) (void *pConnCtx);
And for comparison, here is how nssock.cpp implements it's
location
proc (this is comparison only, note that drv.c doesn't
actually call
into nssock/SockLocation since nssock supplies it's own
thread).
static char *
SockLocation(void *arg)
{
Conn *connPtr = arg;
This affects drivers that use the drv.c/RunDriver, which
nssock
doesn't (and nsunix does.) (Near as I can tell, nsunix is
the only
user of this routine in what used to be core AOLserver,
does anyone
else use drv.c/RunDriver?)
The fix for this is to not call locationProc at this
moment -- it's use is
only to generate a log message.
Claim 2. In the excerpt:
while ((status = ((*dPtr-acceptProc)(dData, cData)))
== NS_OK) {
if (Ns_QueueConn(dData, cData) != NS_OK) {
(*dPtr-closeProc)(dData);
}
}
Ns_QueueConn has been called with the wrong parameters. It
should be called as:
if (Ns_QueueConn(dPtr, cData) != NS_OK) {
in nsd/serv.c we can see how Ns_QueueConn is defined:
int
Ns_QueueConn(void *drvPtr, void *drvData)
The first argument is a pointer to a driver, the second
argument is a
handle to a driver specific data structure (each driver
family sticks
different things in there.)
Claim 3. In the same excerpt, closeProc should be called
with cData
and not dData.
--
You can respond by visiting:
http://sourceforge.net/tracker/?func=detailatid=103152aid=449744group_id=3152
[AOLSERVER] [ aolserver-Patches-456225 ] ns_logdump command
Patches item #456225, was opened at 2001-08-28 11:08 You can respond by visiting: http://sourceforge.net/tracker/?func=detailatid=303152aid=456225group_id=3152 Category: aolserver3_3 Group: None Status: Open Resolution: None Priority: 5 Submitted By: Jerry Asher (jerryasher) Assigned to: Nobody/Anonymous (nobody) Summary: ns_logdump command Initial Comment: Dump a string in hex and ascii to the log. Available from C or Tcl. Useful in debugging new protocols Add a C API and Tcl API for dumping HEX to the log. void Ns_LogDumpString(Ns_LogSeverity severity, char *s, int len) and ns_logdump severity string Severity is a standard ns_log severity constant, Notice, Warning, Error, Debug, or Dev. The output in the log file looks like: : 4164 6420 6120 4320 4150 4920 616e 6420 Add a C API and 0010: 5463 6c20 4150 4920 666f 7220 6475 6d70 Tcl API for dump 0020: 696e 6720 4845 5820 746f 2074 6865 206c ing HEX to the l 0030: 6f67 2e0a 0a76 6f69 6420 4e73 5f4c 6f67 og...void Ns_Log 0040: 4475 6d70 5374 7269 6e67 284e 735f 4c6f DumpString(Ns_Lo The length of a line is controlled by a tcl config parameter, dumpbytesperline, that defaults to 32. ns_section ns/parameters ns_paramdumpbytesperline 32 This patch modifies: nsd/log.c, nsd/nsconf.c, nsd/nsconf.h, nsd/nsd.h, nsd/tclcmds.c. -- You can respond by visiting: http://sourceforge.net/tracker/?func=detailatid=303152aid=456225group_id=3152
[AOLSERVER] [ aolserver-Patches-456225 ] ns_logdump command
Patches item #456225, was opened at 2001-08-28 11:08 You can respond by visiting: http://sourceforge.net/tracker/?func=detailatid=303152aid=456225group_id=3152 Category: aolserver3_3 Group: None Status: Open Resolution: None Priority: 5 Submitted By: Jerry Asher (jerryasher) Assigned to: Nobody/Anonymous (nobody) Summary: ns_logdump command Initial Comment: Dump a string in hex and ascii to the log. Available from C or Tcl. Useful in debugging new protocols Add a C API and Tcl API for dumping HEX to the log. void Ns_LogDumpString(Ns_LogSeverity severity, char *s, int len) and ns_logdump severity string Severity is a standard ns_log severity constant, Notice, Warning, Error, Debug, or Dev. The output in the log file looks like: : 4164 6420 6120 4320 4150 4920 616e 6420 Add a C API and 0010: 5463 6c20 4150 4920 666f 7220 6475 6d70 Tcl API for dump 0020: 696e 6720 4845 5820 746f 2074 6865 206c ing HEX to the l 0030: 6f67 2e0a 0a76 6f69 6420 4e73 5f4c 6f67 og...void Ns_Log 0040: 4475 6d70 5374 7269 6e67 284e 735f 4c6f DumpString(Ns_Lo The length of a line is controlled by a tcl config parameter, dumpbytesperline, that defaults to 32. ns_section ns/parameters ns_paramdumpbytesperline 32 This patch modifies: nsd/log.c, nsd/nsconf.c, nsd/nsconf.h, nsd/nsd.h, nsd/tclcmds.c. -- Comment By: Jerry Asher (jerryasher) Date: 2001-08-28 11:09 Message: Logged In: YES user_id=20647 Uh, the actual format looks a lot better than what sourceforge did to my prior comment. Looks good actually! -- You can respond by visiting: http://sourceforge.net/tracker/?func=detailatid=303152aid=456225group_id=3152
[AOLSERVER] [ aolserver-Bugs-455657 ] ns_db verbose works on handle not pool
Bugs item #455657, was opened at 2001-08-26 20:38 You can respond by visiting: http://sourceforge.net/tracker/?func=detailatid=103152aid=455657group_id=3152 Category: Architecture: Server (nsd) Group: aolserver3_3 Status: Open Resolution: None Priority: 5 Submitted By: Jerry Asher (jerryasher) Assigned to: Nobody/Anonymous (nobody) Summary: ns_db verbose works on handle not pool Initial Comment: The ns_db verbose command operates on a db handle. It doesn't operate on the db pool as the documentation describes. It would be good to have a command that did operate on a db pool in its entirety. -- Comment By: Jerry Asher (jerryasher) Date: 2001-08-27 01:31 Message: Logged In: YES user_id=20647 I've attached a patch that creates an ns_poolverbosity command. ns_poolverbosity ?poolname ?on | ?off It also modifiles nsd/dbinit.c/LogSql to check on the pool fverbose when deciding whether or not to log the sql. -- You can respond by visiting: http://sourceforge.net/tracker/?func=detailatid=103152aid=455657group_id=3152
[AOLSERVER] [ aolserver-Patches-456225 ] ns_logdump command
Patches item #456225, was opened at 2001-08-28 11:08 You can respond by visiting: http://sourceforge.net/tracker/?func=detailatid=303152aid=456225group_id=3152 Category: aolserver3_3 Group: None Status: Open Resolution: None Priority: 5 Submitted By: Jerry Asher (jerryasher) Assigned to: Nobody/Anonymous (nobody) Summary: ns_logdump command Initial Comment: Dump a string in hex and ascii to the log. Available from C or Tcl. Useful in debugging new protocols Add a C API and Tcl API for dumping HEX to the log. void Ns_LogDumpString(Ns_LogSeverity severity, char *s, int len) and ns_logdump severity string Severity is a standard ns_log severity constant, Notice, Warning, Error, Debug, or Dev. The output in the log file looks like: : 4164 6420 6120 4320 4150 4920 616e 6420 Add a C API and 0010: 5463 6c20 4150 4920 666f 7220 6475 6d70 Tcl API for dump 0020: 696e 6720 4845 5820 746f 2074 6865 206c ing HEX to the l 0030: 6f67 2e0a 0a76 6f69 6420 4e73 5f4c 6f67 og...void Ns_Log 0040: 4475 6d70 5374 7269 6e67 284e 735f 4c6f DumpString(Ns_Lo The length of a line is controlled by a tcl config parameter, dumpbytesperline, that defaults to 32. ns_section ns/parameters ns_paramdumpbytesperline 32 This patch modifies: nsd/log.c, nsd/nsconf.c, nsd/nsconf.h, nsd/nsd.h, nsd/tclcmds.c. -- Comment By: Jerry Asher (jerryasher) Date: 2001-08-28 12:39 Message: Logged In: YES user_id=20647 Use the second patch, not the first. The first patch doesn't declare Ns_LogDumpString within include/ns.h. (Nb: I tried deleting the patch, but sourceforge would not let me.) -- Comment By: Jerry Asher (jerryasher) Date: 2001-08-28 11:09 Message: Logged In: YES user_id=20647 Uh, the actual format looks a lot better than what sourceforge did to my prior comment. Looks good actually! -- You can respond by visiting: http://sourceforge.net/tracker/?func=detailatid=303152aid=456225group_id=3152
[AOLSERVER] [ aolserver-Bugs-449744 ] conn driver freeproc never called
Bugs item #449744, was opened at 2001-08-10 02:02
You can respond by visiting:
http://sourceforge.net/tracker/?func=detailatid=103152aid=449744group_id=3152
Category: Architecture: Server (nsd)
Group: None
Status: Open
Resolution: None
Priority: 5
Submitted By: Jerry Asher (jerryasher)
Assigned to: Nobody/Anonymous (nobody)
Summary: conn driver freeproc never called
Initial Comment:
Communications drivers can have an Ns_ConnFreeProc.
Although it is not quite clear what this procedure
does, the legacy nsunix code choose to use it to free
various buffers, specifically, the conn input buffer.
But no one ever calls these freeing procedures.
My suggestion: define a routine similar to
conn.c/Ns_ConnClose (called Ns_ConnFree...) that calls
connPtr-drvPtr-freeProc (if it's not null).
Where to call it though? I am not certain.
I chose to call it in in serv.c/ConnRun, as the second
to last statement, right after NsRunCleanups and right
before the dstring is freed.
Is this the right place?
--
Comment By: Jerry Asher (jerryasher)
Date: 2001-08-27 08:20
Message:
Logged In: YES
user_id=20647
This patch contains fixes for four bugs:
1. Memory leak by any communication driver that implements
a driver
free proc: the bug is that this freeproc is never
called within
AOLserver.
Create conn.c/Ns_ConnFree to provide a connection a way
to call
the driver free proc (supported in registration of a
driver, but
never called within AOLserver.)
Call Ns_ConnFree in serv.c/ConnRun at end of
connection. Also
call Ns_ConnFree within drv.c whenever there is a
problem queueing
connection.
Declare Ns_ConnFree within ns.h
2. in nsd/drv.c/RunDriver, there are three bugs present in
which well
defined functions are called with the wrong
structures. (Or
called before the structures have been initialized.)
Call it the
curse of casting, right? No compiler warnings as the
*wrong*
structures are correctly cast into the right types.
To see this, the best way is to compare how similar
routines are
called or defined by the nssock driver.
Claim 1: In the excerpt
dPtr = arg;
dData = dPtr-drvData;
if (dPtr-locationProc != NULL) {
loc = (*dPtr-locationProc)(dData);
. . .
while ((status = ((*dPtr-acceptProc)(dData, cData)))
== NS_OK) {
A locationProc is supposed to take a connection structure,
if so, then
locationProc is being called before the accept, and so the
connection
data (presumably dData) has not yet been initialized.
(I'll make the
claim that dData is not necessarily the connection
structure that
locationproc needs, but regardless of whether it is or
isn't, it's not
fair to ask locationproc to crawl down the connection
structure before
the connection has been accepted.)
Here is how ns.h defines the locationProc:
typedef char *(Ns_ConnLocationProc) (void *pConnCtx);
And for comparison, here is how nssock.cpp implements it's
location
proc (this is comparison only, note that drv.c doesn't
actually call
into nssock/SockLocation since nssock supplies it's own
thread).
static char *
SockLocation(void *arg)
{
Conn *connPtr = arg;
This affects drivers that use the drv.c/RunDriver, which
nssock
doesn't (and nsunix does.) (Near as I can tell, nsunix is
the only
user of this routine in what used to be core AOLserver,
does anyone
else use drv.c/RunDriver?)
The fix for this is to not call locationProc at this
moment -- it's use is
only to generate a log message.
Claim 2. In the excerpt:
while ((status = ((*dPtr-acceptProc)(dData, cData)))
== NS_OK) {
if (Ns_QueueConn(dData, cData) != NS_OK) {
(*dPtr-closeProc)(dData);
}
}
Ns_QueueConn has been called with the wrong parameters. It
should be called as:
if (Ns_QueueConn(dPtr, cData) != NS_OK) {
in nsd/serv.c we can see how Ns_QueueConn is defined:
int
Ns_QueueConn(void *drvPtr, void *drvData)
The first argument is a pointer to a driver, the second
argument is a
handle to a driver specific data structure (each driver
family sticks
different things in there.)
Claim 3. In the same excerpt, closeProc should be called
with cData
and not dData.
--
You can respond by visiting:
http://sourceforge.net/tracker/?func=detailatid=103152aid=449744group_id=3152
[AOLSERVER] [ aolserver-Patches-456052 ] support html4 b.2.2, semicolons in uris
Patches item #456052, was opened at 2001-08-28 00:56 You can respond by visiting: http://sourceforge.net/tracker/?func=detailatid=303152aid=456052group_id=3152 Category: aolserver3_3 Group: None Status: Open Resolution: None Priority: 5 Submitted By: Jerry Asher (jerryasher) Assigned to: Nobody/Anonymous (nobody) Summary: support html4 b.2.2, semicolons in uris Initial Comment: Patch to support HTML4.0 specification Appendix B.2.2, regarding special characters in URL attribute values, specifically using a semicolon, ';', in a URI instead of an ampersand, ''. With this patch, AOLserver will take a URI that uses semicolons and not ampersands and correctly parse the URI into form variables and their values. quote http://www.w3.org/TR/REC-html40/appendix/notes.html#h- B.2.2 B.2.2 Ampersands in URI attribute values The URI that is constructed when a form is submitted may be used as an anchor-style link (e.g., the href attribute for the A element). Unfortunately, the use of the character to separate form fields interacts with its use in SGML attribute values to delimit character entity references. For example, to use the URI http://host/?x=1y=2; as a linking URI, it must be written A href=http://host/?x=1#38;y=2; or A href=http://host/?x=1amp;y=2;. We recommend that HTTP server implementors, and in particular, CGI implementors support the use of ; in place of to save authors the trouble of escaping characters in this manner. /quote This patch adds a config.tcl parameter ns_section ns/server/yourserver ns_param semicolonqueryseparator yes The default value is no, or not examining URI's for semicolons. nsd/nsconf.c - read ns_param semicolonseparator for ns_querytoset nsd/nsconf.h - define CONN_SEMICOLON_SEP_BOOL for ns_querytoset nsd/nsd.h - added semicolonseparator to struct conn in struct _nsconf for ns_querytoset nsd/conn.c - NS_QueryToSet - modified to examine semicolons inside of a request -- You can respond by visiting: http://sourceforge.net/tracker/?func=detailatid=303152aid=456052group_id=3152
Re: [AOLSERVER] /NS/Admin forms for 3.x
Daniel Stasinski did a package.. you can find it at www.scriptkitties.com Its well worth the moments it takes to download and install -- Patrick Spence [EMAIL PROTECTED] Want weather reports delevered to your desktop? http://ww2.weatherbug.com/aff/default.asp?ZCode=z3345 - Original Message - From: James Ranson [EMAIL PROTECTED] To: [EMAIL PROTECTED] Sent: Friday, August 24, 2001 3:39 PM Subject: [AOLSERVER] /NS/Admin forms for 3.x We recently had a new Web Server setup and the Webhost installed AOLServer 3.2, which does not include the NS/Admin forms (I'm a die-hard 2.3.3 user). I recall reading somewhere that a third-party had developed a utility that could accomplish much of what the NS/Admin forms did. Anyone know where I could get a copy of this?
[AOLSERVER] AOLserver mailing list suspended and now freed
The AOLserver mailing list was suspended for a few days due to excessive postings--the threshold of 50 messages per day was reached. I have freed the list. Thanks to Jerry Asher for pinging me about this. Kris
Re: [AOLSERVER] member site charging
Patrick Spence wrote: Those of you who were interested in my experiences with setting up ccBill to handle credit card charging for a member based site with Aolserver please email be offlist at [EMAIL PROTECTED] as I am in the process of writing up my description of what I did, and collecting all the resources in one location.. this way I can give those interested a full description of what I did and how I got it to work... Enjoy... Patrick i'd be interested in your experiences. please send me whatever info you distribute. thanks.. -mike hoegeman, [EMAIL PROTECTED]
Re: [AOLSERVER] Question about extending AOLserver in C
+-- On Aug 23, Michael Roberts said: So I'm studying set.c and tclset.c, because the ns_set is probably the extension I use the most and I figure it's Done Right -- but I can't for the life of me see how the connection-local interpreter knows to call Ns_TclFreeSet. Where is the proper cleanup function assigned to the interpreter? I'm so confused If you execute ns_set free $setid, then NsTclSetCmd will call Ns_TclFreeSet. Sets that you don't delete this way will be deleted at interpreter cleanup time by FreeSets.
