Re: memory leaks in orbit

[ABHISHEK THAKUR <abhishekthakur usa net>]

Hi,
   I am sorry if my repetitive mails are causing a concern .
   I have continued with my debugging and this is what I fould -
"resolve" request for orbit-name-server does not leak memory, but the "list"
request does. And the memory seems to be leaking in "CosNaming-common.c".
   While I continue with the hunt - i remember some issues of memeory leak in
"common" code w.r.t. structs/sequences being discussed. What i observed may be
part of the same - but I am not sure(and hence a cc goes to
andrewm treshna com )
   I also remember coming accross a posting on use of "free_strings" and/or
"val->_release". Though at that time I could not understand the exact issues
being discussed - i am not sure if they are bugging me.

   From the gurus I would like to get the following clerification -

in "new-name-server.c" line 707 of 793 (more specifically "new-name-server.c"
line 661 of 793) - 'bl' is being allocated using list creation operator. who
is responsible to free it?

shouldn't the '_release' be set on this somewhere so that the array is freed 
after the value is sent to the client. 
As I see the CORBA_sequence_CosNaming_Binding__free is being invoked through
'CORBA_free(bl);' in *skel.c but in there _release is not set and hence the
value is not released.

my debugging shows that this might be the possible cause for the leak.


thanks for any pointers.

Abhishek

Abhishek Thakur    @
Software Engineer
Synchronous Inc.
Ph:(o)408-284-6136 (r)408-730-1690 (fx)815-377-0514

---

Get free email and a permanent address at http://www.netaddress.com