Re: [PATCH] more precise throughput using double [v3]

Vincent CHAVANIS schrieb:
> great !
> 
> One more thing in Changelog
> 
>>    that changes throughput from interger
>                                                      ^^ should read integer

fixed as well, thanks ;)

Thanks,
Alex

> 
> regards
> 
> Vincent
> 
> --
> Telemaque - NICE - (FR)
> Service Technique - Developpement
> http://www.telemaque.fr/
> [email protected]
> Tel : +33 4 93 97 71 64 (fax 68)
> 
> ----- Original Message ----- 
> From: "Alexander Malysh" <[email protected]>
> To: <[email protected]>
> Sent: Friday, April 07, 2006 6:37 PM
> Subject: Re: [PATCH] more precise throughput using double [v3]
> 
> 
>> oops, you are right... this patch committed to cvs:
>>
>> Index: gw/smscconn.c
>> ===================================================================
>> RCS file: /home/cvs/gateway/gw/smscconn.c,v
>> retrieving revision 1.54
>> diff -a -u -p -r1.54 smscconn.c
>> --- gw/smscconn.c       7 Apr 2006 12:49:50 -0000       1.54
>> +++ gw/smscconn.c       7 Apr 2006 16:35:33 -0000
>> @@ -151,7 +151,6 @@ SMSCConn *smscconn_create(CfgGroup *grp,
>>      SMSCConn *conn;
>>      Octstr *smsc_type;
>>      int ret;
>> -    double throughput;
>>      Octstr *allowed_smsc_id_regex;
>>      Octstr *denied_smsc_id_regex;
>>      Octstr *allowed_prefix_regex;
>> @@ -218,10 +217,10 @@ SMSCConn *smscconn_create(CfgGroup *grp,
>>              panic(0, "Could not compile pattern '%s'", 
>> octstr_get_cstr(preferred_prefix_regex));
>>
>>      if ((tmp = cfg_get(grp, octstr_imm("throughput"))) != NULL) {
>> -        if (octstr_parse_double(&throughput, tmp, 0) == -1)
>> +        if (octstr_parse_double(&conn->throughput, tmp, 0) == -1)
>>              conn->throughput = 0;
>>          octstr_destroy(tmp);
>> -        info(0, "Set throughput to %.3f for smsc id <%s>", throughput, 
>> octstr_get_cstr(conn->id));
>> +        info(0, "Set throughput to %.3f for smsc id <%s>", 
>> conn->throughput, octstr_get_cstr(conn->id));
>>      }
>>
>>      /* configure the internal rerouting rules for this smsc id */
>>
>>
>> Thanks,
>> Alex
>>
>> Vincent CHAVANIS schrieb:
>>> wow, something is going wrong with your modifications.
>>>
>>> Here is the fix, 
>>>
>>> --- /gateway-cvs/gw/smscconn.c  2006-04-07 14:54:05.000000000 +0200
>>> +++ /gateway/gw/smscconn.c      2006-04-07 16:54:44.000000000 +0200
>>> @@ -220,8 +220,10 @@
>>>      if ((tmp = cfg_get(grp, octstr_imm("throughput"))) != NULL) {
>>>          if (octstr_parse_double(&throughput, tmp, 0) == -1)
>>>              conn->throughput = 0;
>>> +        else
>>> +            conn->throughput = (double) throughput;
>>>          octstr_destroy(tmp);
>>> -        info(0, "Set throughput to %.3f for smsc id <%s>", throughput, octstr_get_cstr(conn->id));
>>> +        info(0, "Set throughput to %.3f for smsc id <%s>", conn->throughput, octstr_get_cstr(conn->id));
>>>      }
>>>
>>> Vincent
>>>
>>> --
>>> Telemaque - NICE - (FR)
>>> Service Technique - Developpement
>>> http://www.telemaque.fr/
>>> [email protected]
>>> Tel : +33 4 93 97 71 64 (fax 68)
>>>
>>> ----- Original Message ----- 
>>> From: "Alexander Malysh" <[email protected]>
>>> To: <[email protected]>
>>> Sent: Friday, April 07, 2006 4:32 PM
>>> Subject: Re: [PATCH] more precise throughput using double [v3]
>>>
>>>
>>>> Vincent CHAVANIS schrieb:
>>>>> Thanks++,
>>>>>
>>>>> I did not see that point.
>>>> cfg_get(...) returns Octstr* that must be destroyed afterwards. and we 
>>>> avoid using strXXX functions if we have the same functionality in Octstr.
>>>>
>>>> Thanks,
>>>> Alex
>>>>
>>>>> Vincent
>>>>>
>>>>> --
>>>>> Telemaque - NICE - (FR)
>>>>> Service Technique - Developpement
>>>>> http://www.telemaque.fr/
>>>>> [email protected]
>>>>> Tel : +33 4 93 97 71 64 (fax 68)
>>>>>
>>>>> ----- Original Message ----- 
>>>>> From: "Alexander Malysh" <[email protected]>
>>>>> To: <[email protected]>
>>>>> Sent: Friday, April 07, 2006 2:51 PM
>>>>> Subject: Re: [PATCH] more precise throughput using double [v3]
>>>>>
>>>>>
>>>>>> Hi,
>>>>>>
>>>>>> this patch applies ok and committed to cvs in modified form.
>>>>>>
>>>>>>>> +    throughput = strtod(octstr_get_cstr(cfg_get(grp, 
>>>>>> octstr_imm("throughput"))),(char **) NULL);
>>>>>>>> +    if (throughput == ERANGE || throughput <= 0)
>>>>>>>>          conn->throughput = 0;   /* defaults to no throughtput
>>>>>> ...
>>>>>>
>>>>>> Here you have memleak. Look into committed version where it's fixed.
>>>>>>
>>>>>> Thanks,
>>>>>> Alex
>>>>>>
>>>>>> Vincent CHAVANIS schrieb:
>>>>>>> is this new one is ok ?
>>>>>>>
>>>>>>> --
>>>>>>> Telemaque - NICE - (FR)
>>>>>>> Service Technique - Developpement
>>>>>>> http://www.telemaque.fr/
>>>>>>> [email protected]
>>>>>>> Tel : +33 4 93 97 71 64 (fax 68)
>>>>>>>
>>>>>>> ----- Original Message ----- 
>>>>>>> From: "Alexander Malysh" <[email protected]>
>>>>>>> To: <[email protected]>
>>>>>>> Sent: Thursday, April 06, 2006 7:23 PM
>>>>>>> Subject: Re: [PATCH] more precise throughput using double [v3]
>>>>>>>
>>>>>>>
>>>>>>>> Hi,
>>>>>>>>
>>>>>>>> you patch doesn't apply. Please redo your patch with cvs -Nau from 
>>>>>>>> within gateway dir.
>>>>>>>>
>>>>>>>> Thanks,
>>>>>>>> Alex
>>>>>>>>
>>>>>>>> ------------------------------------------------------------------------
>>>>>>>>
>>>>>>>> diff -i -E -b -w -B -rNau gateway/gw/smsc/smsc_emi.c gateway-patch/gw/smsc/smsc_emi.c
>>>>>>>> --- gateway/gw/smsc/smsc_emi.c  2005-02-11 16:35:48.000000000 +0100
>>>>>>>> +++ gateway-patch/gw/smsc/smsc_emi.c    2006-04-06 19:42:24.000000000 +0200
>>>>>>>> @@ -992,7 +992,7 @@
>>>>>>>>      Msg *msg;
>>>>>>>>      double delay = 0;
>>>>>>>>  
>>>>>>>> -    if (conn->throughput) {
>>>>>>>> +    if (conn->throughput > 0) {
>>>>>>>>          delay = 1.0 / conn->throughput;
>>>>>>>>      }
>>>>>>>>      
>>>>>>>> @@ -1001,7 +1001,7 @@
>>>>>>>>             (msg = gw_prioqueue_remove(PRIVDATA(conn)->outgoing_queue)) != NULL) {
>>>>>>>>          int nexttrn = emi2_next_trn(conn);
>>>>>>>>  
>>>>>>>> -        if (conn->throughput)
>>>>>>>> +        if (conn->throughput > 0)
>>>>>>>>              gwthread_sleep(delay);
>>>>>>>>  
>>>>>>>>          /* convert the generic Kannel message into an EMI type message */
>>>>>>>> diff -i -E -b -w -B -rNau gateway/gw/smsc/smsc_fake.c gateway-patch/gw/smsc/smsc_fake.c
>>>>>>>> --- gateway/gw/smsc/smsc_fake.c 2005-02-11 16:35:48.000000000 +0100
>>>>>>>> +++ gateway-patch/gw/smsc/smsc_fake.c   2006-04-06 19:42:38.000000000 +0200
>>>>>>>> @@ -243,7 +243,7 @@
>>>>>>>>      Msg        *msg;
>>>>>>>>      double delay = 0;
>>>>>>>>  
>>>>>>>> -    if (conn->throughput) {
>>>>>>>> +    if (conn->throughput > 0) {
>>>>>>>>          delay = 1.0 / conn->throughput;
>>>>>>>>      }
>>>>>>>>  
>>>>>>>> @@ -297,7 +297,7 @@
>>>>>>>>              }
>>>>>>>>  
>>>>>>>>              /* obey throughput speed limit, if any */
>>>>>>>> -            if (conn->throughput) {
>>>>>>>> +            if (conn->throughput > 0) {
>>>>>>>>                  gwthread_sleep(delay);
>>>>>>>>              }
>>>>>>>>          }
>>>>>>>> diff -i -E -b -w -B -rNau gateway/gw/smsc/smsc_http.c gateway-patch/gw/smsc/smsc_http.c
>>>>>>>> --- gateway/gw/smsc/smsc_http.c 2006-03-29 11:17:55.000000000 +0200
>>>>>>>> +++ gateway-patch/gw/smsc/smsc_http.c   2006-04-06 19:43:00.000000000 +0200
>>>>>>>> @@ -1308,7 +1308,7 @@
>>>>>>>>      Msg *sms = msg_duplicate(msg);
>>>>>>>>      double delay = 0;
>>>>>>>>  
>>>>>>>> -    if (conn->throughput) {
>>>>>>>> +    if (conn->throughput > 0) {
>>>>>>>>          delay = 1.0 / conn->throughput;
>>>>>>>>      }
>>>>>>>>  
>>>>>>>> @@ -1316,7 +1316,7 @@
>>>>>>>>      conndata->send_sms(conn, sms);
>>>>>>>>  
>>>>>>>>      /* obey throughput speed limit, if any */
>>>>>>>> -    if (conn->throughput)
>>>>>>>> +    if (conn->throughput > 0)
>>>>>>>>          gwthread_sleep(delay);
>>>>>>>>  
>>>>>>>>      return 0;
>>>>>>>> diff -i -E -b -w -B -rNau gateway/gw/smsc/smsc_smasi.c gateway-patch/gw/smsc/smsc_smasi.c
>>>>>>>> --- gateway/gw/smsc/smsc_smasi.c        2005-02-11 16:35:48.000000000 +0100
>>>>>>>> +++ gateway-patch/gw/smsc/smsc_smasi.c  2006-04-06 19:43:19.000000000 +0200
>>>>>>>> @@ -872,7 +872,7 @@
>>>>>>>>  
>>>>>>>>      if (*pending_submits == -1) return;
>>>>>>>>  
>>>>>>>> -    if (smasi->conn->throughput) {
>>>>>>>> +    if (smasi->conn->throughput > 0) {
>>>>>>>>          delay = 1.0 / smasi->conn->throughput;
>>>>>>>>      }
>>>>>>>>  
>>>>>>>> @@ -894,7 +894,7 @@
>>>>>>>>          smasi_pdu_destroy(pdu);
>>>>>>>>  
>>>>>>>>          /* obey throughput speed limit, if any */
>>>>>>>> -        if (smasi->conn->throughput)
>>>>>>>> +        if (smasi->conn->throughput > 0)
>>>>>>>>              gwthread_sleep(delay);
>>>>>>>>  
>>>>>>>>          ++(*pending_submits);
>>>>>>>> diff -i -E -b -w -B -rNau gateway/gw/smsc/smsc_smpp.c gateway-patch/gw/smsc/smsc_smpp.c
>>>>>>>> --- gateway/gw/smsc/smsc_smpp.c 2006-02-07 15:54:44.000000000 +0100
>>>>>>>> +++ gateway-patch/gw/smsc/smsc_smpp.c   2006-04-06 19:44:14.000000000 +0200
>>>>>>>> @@ -1013,7 +1013,7 @@
>>>>>>>>              /*
>>>>>>>>               * obey throughput speed limit, if any.
>>>>>>>>               */
>>>>>>>> -            if (smpp->conn->throughput)
>>>>>>>> +            if (smpp->conn->throughput > 0)
>>>>>>>>                  gwthread_sleep(delay);
>>>>>>>>          }
>>>>>>>>          else { /* write error occurs */
>>>>>>>> diff -i -E -b -w -B -rNau gateway/gw/smscconn.c gateway-patch/gw/smscconn.c
>>>>>>>> --- gateway/gw/smscconn.c       2006-04-01 19:48:51.000000000 +0200
>>>>>>>> +++ gateway-patch/gw/smscconn.c 2006-04-06 19:51:57.000000000 +0200
>>>>>>>> @@ -64,6 +64,7 @@
>>>>>>>>  
>>>>>>>>  #include <signal.h>
>>>>>>>>  #include <time.h>
>>>>>>>> +#include <errno.h>
>>>>>>>>  
>>>>>>>>  #include "gwlib/gwlib.h"
>>>>>>>>  #include "gwlib/regex.h"
>>>>>>>> @@ -151,7 +152,7 @@
>>>>>>>>      SMSCConn *conn;
>>>>>>>>      Octstr *smsc_type;
>>>>>>>>      int ret;
>>>>>>>> -    long throughput;
>>>>>>>> +    double throughput;
>>>>>>>>      Octstr *allowed_smsc_id_regex;
>>>>>>>>      Octstr *denied_smsc_id_regex;
>>>>>>>>      Octstr *allowed_prefix_regex;
>>>>>>>> @@ -216,10 +217,13 @@
>>>>>>>>          if ((conn->preferred_prefix_regex = gw_regex_comp(preferred_prefix_regex, REG_EXTENDED)) == NULL)
>>>>>>>>              panic(0, "Could not compile pattern '%s'", octstr_get_cstr(preferred_prefix_regex));
>>>>>>>>  
>>>>>>>> -    if (cfg_get_integer(&throughput, grp, octstr_imm("throughput")) == -1)
>>>>>>>> +    throughput = strtod(octstr_get_cstr(cfg_get(grp, octstr_imm("throughput"))),(char **) NULL);
>>>>>>>> +    if (throughput == ERANGE || throughput <= 0)
>>>>>>>>          conn->throughput = 0;   /* defaults to no throughtput limitation */
>>>>>>>>      else
>>>>>>>> -        conn->throughput = (int) throughput;
>>>>>>>> +        conn->throughput = throughput;
>>>>>>>> +
>>>>>>>> +    info(0, "Set throughput to %.3f for smsc id <%s>", throughput, octstr_get_cstr(conn->id));
>>>>>>>>  
>>>>>>>>      /* configure the internal rerouting rules for this smsc id */
>>>>>>>>      init_reroute(conn, grp);
>>>>>>>> diff -i -E -b -w -B -rNau gateway/gw/smscconn_p.h gateway-patch/gw/smscconn_p.h
>>>>>>>> --- gateway/gw/smscconn_p.h     2005-02-11 16:35:48.000000000 +0100
>>>>>>>> +++ gateway-patch/gw/smscconn_p.h       2006-04-06 19:46:51.000000000 +0200
>>>>>>>> @@ -194,7 +194,7 @@
>>>>>>>>  
>>>>>>>>      int alt_dcs; /* use alternate DCS 0xFX */
>>>>>>>>  
>>>>>>>> -    int throughput;     /* message thoughput per sec. to be delivered to SMSC */
>>>>>>>> +    double throughput;     /* message thoughput per sec. to be delivered to SMSC */
>>>>>>>>  
>>>>>>>>      /* Stores rerouting information for this specific smsc-id */
>>>>>>>>      int reroute;                /* simply turn MO into MT and process internally */
>>>>>
>>>>
>>>
>>>
>>
>>
> 
> 
>