ayatana-commits team mailing list archive

[Aurélien Gâteau <aurelien.gateau@xxxxxxxxxxxxx>]

>   review needs-information
> 
> The remove filter makes complete sense to me.
> 
> The return on the destruction of the proxy struct doesn't.  It seems
> like if the proxy is already freed, we'd still need to clean up the
> members of the proxy_t object itself.  It seems like that'd just
> introduce a memory leak as if the proxy dies we'd never destroy the
> struct.

I assumed this line from proxy_destroyed():

	proxyt->proxy = NULL; /* Clear this so we don't get a double free on this guy */

was setting proxy to NULL as a marker to avoid a crash in proxy_struct_destroy(), but now that I think more about it, I am probably wrong. My first attempt was to modify proxy_destroyed() so that it would remove the proxy_t instance from the listener proxies list. It failed, but it made more sense I think.
-- 
https://code.launchpad.net/~agateau/libindicate/free-fixes/+merge/18996
Your team ayatana-commits is subscribed to branch lp:libindicate.