[Pidgin] #8786: small memory leaks
Pidgin
trac at pidgin.im
Thu Mar 26 01:14:05 EDT 2009
#8786: small memory leaks
------------------------------------+---------------------------------------
Reporter: darkrain42 | Owner:
Type: patch | Status: new
Milestone: Patches Needing Review | Component: libpurple
Version: 2.5.5 | Resolution:
Keywords: | Launchpad_bug:
------------------------------------+---------------------------------------
Comment(by darkrain42):
Replying to [comment:2 datallah]:
> I don't think the (first) patch is actually correct; the list items are
not actually copied anywhere, so in the cases where the value you're
adding is a pointer to a static buffer as it would be from any of the
_(...) gettext functions, the value will end up changing on you. I'm not
sure what the status of the SILC ciphers and hmacs arrays is in terms of
expectation of a availability.
Replying to [comment:2 datallah]:
> I don't think the (first) patch is actually correct; the list items are
not actually copied anywhere, so in the cases where the value you're
adding is a pointer to a static buffer as it would be from any of the
_(...) gettext functions, the value will end up changing on you. I'm not
sure what the status of the SILC ciphers and hmacs arrays is in terms of
expectation of a availability.
I was under the impression (although I don't know where I got it) that _()
returns the string in a stable location, as it just returns a pointer to
string in whatever structure NLS stores them in after they're read in from
memory.
The silc arrays are declared statically (and const) in libsilc.
> Technically these aren't actually leaks as such unless libpurple is
expected to clean up everything during uninit() - they are just long-
living data that isn't ever freed (there is always a reference to it).
Yeah, I realize that; they're minor enough that it's not that big of a
deal.
--
Ticket URL: <http://developer.pidgin.im/ticket/8786#comment:3>
Pidgin <http://pidgin.im>
Pidgin
More information about the Tracker
mailing list