BTW, while I'm adding the meta_data field on all DB engines, and I've
noticed a difference on dlr_oracle.c, on dlr_get_oracle:
sql = octstr_format("SELECT %S, %S, %S, %S, %S, %S, %S FROM %S
WHERE %S=:1 AND %S=:2 AND %S=:3 AND ROWNUM < 2",
fields->field_mask, fields->field_serv,
fields->field_url, fields->field_src,
fields->field_dst, fields->field_boxc,
fields->table, fields->field_smsc,
fields->field_ts, fields->field_dst);
While on all other engines, field_dst is not used on the condition:
pgsql:
sql = octstr_format("SELECT %s, %s, %s, %s, %s, %s FROM %s WHERE
%s='%s' AND %s='%s' LIMIT 1;",
sdb:
sql = octstr_format("SELECT %s, %s, %s, %s, %s, %s FROM %s WHERE
%s='%s' AND %s='%s' %s", [NO LIMIT CLAUSE]
mysql:
sql = octstr_format("SELECT %S, %S, %S, %S, %S, %S FROM %S WHERE %S=?
AND %S=? LIMIT 1",
and so on.
Any reason why this is done that way? Which one should be fixed?
Regards,
--
Alejandro Guerrieri
[email protected]
On 04/09/2009, at 13:46, Alexander Malysh wrote:
Am 04.09.2009 um 13:05 schrieb Alejandro Guerrieri:
2009-09-04 Alexander Malysh <amalysh at kannel.org>
* aclocal.m4, configure, configure.in, gw/dlr_mysql.c, gwlib/
dbpool_mysql.c,
test/test_dbpool.c: implemented select/update function calls
for dbpool_mysql. This
require mysql prepaid statement support and therefore mysql
version >= 4.1.
Changed dlr_mysql to use these new function calls. This fixes
#258.
Would make sense if I update sqlbox to use those calls as well right?
yep
Regards,
--
Alejandro Guerrieri
[email protected]
On 04/09/2009, at 9:45, Alexander Malysh wrote:
ahh alex,
before you start to add meta_data to db storages please give me
time to merge mysql-prepaid branch. Just update your cvs in
30-60min.
Thanks,
Alexander Malysh
Am 04.09.2009 um 09:40 schrieb Alexander Malysh:
Hi Alex,
if you add:
+ Octstr *meta_data;
to dlr struct then you have add meta_data column to all storage
types :) otherwise don't add it. If you ask me, I would say, if
you add it for faked dlrs then you have
to add it to db storages as well.
as to the points:
1) +0
2) -1 because I don't see reason to add internal smpp sequence.
this is just one counter and will not add any useful information.
3) ++1 :)
Thanks,
Alexander Malysh
Am 04.09.2009 um 01:08 schrieb Alejandro Guerrieri:
Ok, silly diff errors aside (I wasn't including dlr_p.h :P),
here's the new patches also changing the error code format as
Alex suggested.
<kannel-dlr-meta-data.diff>
<kannel-dlr-command-status.diff>
Userguide patch coming later.
Just a few more ideas to consider:
1. Adding one more called "dlr_status_text" with the text
description from smpp_error_to_string(pdu-
>u.submit_sm_resp.command_status). Not sure if it makes sense,
this could be done at the application level with a simple table...
2. Adding one more called "dlr_sequence" with the
sequence_number parameter.
3. Extending patch #1 adding the meta-data column on all DB
implementations, so meta-data could be passed and stored on
dlr's other than the first one generated by kannel.
Regards,
--
Alejandro Guerrieri
[email protected]
On 03/09/2009, at 23:33, Alexander Malysh wrote:
Hi Alex,
first patch seems still broken:
+ dlr->meta_data = (msg->sms.meta_data ?
octstr_duplicate(msg->sms.meta_data) : octstr_create(""));
we don't have meta_data in dlr struct. otherwise +1
second patch:
+ if (msg->sms.meta_data == NULL)
+ msg->sms.meta_data = octstr_create("");
+ meta_data_set_value(msg->sms.meta_data, "smpp",
octstr_imm("dlr_status"),
+ octstr_format("%d", pdu-
>u.submit_sm_resp.command_status), 1);
+
Would it it not better to have error code formated as smpp-tlv
group and as error message?
+ octstr_format("0x%08lx", pdu-
>u.submit_sm_resp.command_status), 1);
Thanks,
Alexander Malysh
Am 03.09.2009 um 21:13 schrieb Alejandro Guerrieri:
Oh, I see, I've attached the ~ backup file from vim. This are
the right files, just in case:
<kannel-dlr-command-status.diff>
<kannel-dlr-meta-data.diff>
--
Alejandro Guerrieri
[email protected]
On 03/09/2009, at 21:06, Alejandro Guerrieri wrote:
Oh, that must have filtered into the patch. Weird :P
Sorry, it's safe to remove it.
Regards,
--
Alejandro Guerrieri
[email protected]
On 03/09/2009, at 20:56, Gustavo Mohme C. wrote:
Hi Alejandro,
I get this error after applying your patch.
gw/dlr.c: In function ‘dlr_add’:
gw/dlr.c:352: error: ‘struct dlr_entry’ has no member named
‘meta_data’
make: *** [gw/dlr.o] Error 1
After editing /gw/dlr_p.h and adding meta_data to the struct
dlr_entry, it compiled with no complaints. Is this the
correct thing to do?
I was patching the latest kannel snapshot....
Thanks,
Gustavo
On Thu, Sep 3, 2009 at 3:39 AM, Alejandro Guerrieri <[email protected]
> wrote:
Hi,
This is a set of two patches, though they're both very
simple and could have been mixed, I've preferred to split to
honor the rule of not mixing features in a single patch.
Patch #1, kannel-drl-meta-data.diff allows meta-data to be
passed when sending a message to come back into the "fake"
dlr that kannel generates when the smsc accepts or rejects a
message. It could be extended to work with the different dlr
engines so the meta-data would be also passed on the
intermediate and final dlr's delivered by the carriers
(though it would require changes on the database structure
to add a "meta_data" column of course. I'm not sure this
would be needed, the dlr-url could be easily abused to pass
application-specific parameters without much hassle.
Patch #2, kannel-dlr-command-status.diff builds on top of
patch #1 and creates a meta data value called "dlr_status"
that comes back on the "fake" dlr generated by kannel (This
was the real reason why patch #1 was created). The value
there is the "command_status" value some people on the list
needs to get from the submit_sm_resp PDU's.
Example usage:
On sendsms:
http://localhost:13013/cgi-bin/sendsms?username=kannel&password=kannel&from=12345&to=12345678&smsc=mysmsc&text=Hello&dlr-mask=31&meta-data=%3Fsmpp%3Fmy_own_field%3D1234&dlr-url=http%3A%2F%2Flocalhost%2Fx%3Fdata%3D%25D
Notes:
meta-data is urlencoded version of: ?smpp?my_own_field=1234
dlr-url is urlencoded version of: http://localhost/x?data=%D
So, after applying Patch #1, kannel would call the following
url:
http://localhost/x?md=%3Fsmpp%3Fmy_own_field%3D1234
The "md" parameter, once urldecoded would look like:
?smpp?my_own_field=1234
You can pass as many parameters as you want of course and
they would be added to meta-data along with any other fields
you've defined on your smpp-tlv groups, etc.
So far so good, in fact you could pass this kind of stuff on
regular url variables, but the goal of this patch was to
allow to add stuff back from the smsc's response.
Now, with patch #2 also applied, the url returned would be
something like this:
http://localhost/x?md=%3Fsmpp%3Fmy_own_field%3D1234%26dlr_status%3D69%26
The "md" parameter, once urldecoded would look like:
?smpp?my_own_field=1234&dlr_status=69&
In this case, dlr_status gets loaded with submit_sm_resp's
command_status which was: 69 = 0x00000045 (Submit Failed).
The "sequence_number" could be added just as easily (not a
bad idea imho). What do you think? "dlr_sequence" or
"dlr_seq" would be ok?
I'm writing the userguide patch if this goes forward of
course.
Comments?
Regards,
--
Alejandro Guerrieri
[email protected]
