the sorry saga of the talloc soname 'fix'
sam at liddicott.com
Wed Jul 8 08:44:34 GMT 2009
* ronnie sahlberg wrote, On 08/07/09 09:27:
>> Before I explain, some points must be acknowledged:
>> 1. There exists much code that uses talloc_steal/talloc_free
>> 2. This code needs fixing anyway.
>> These are the features of the solution:
>> 1. To make talloc_steal and talloc_free unambiguous
> There is nothing inherently ambigous with talloc_steal and talloc_free.
> I use them a lot in CTDB and there is aboslutely no need to fix
> anything regarding these functions or their use there.
That's OK, cos I wasn't fixing anything regarding their use there, and
the fix I suggested won't interfere in the slightest with the use there.
> These functions are NOT ambiguous by themself. They are only ambiguous
> when they are used together with _reference/unreference. Never else.
The bit when you said "they are only ambiguous when...." - that's what
we're talking about. I'm glad that you recognize where the ambiguity exists.
I spent a good amount of time trying to put together a post where people
would consider the suggestion and not get side-tracked.
I find it funny when you effectively say "ah, but I can think of a use
when they are not ambiguous" on a thread where we are talking about the
use when they are ambiguous.
Now that we understood that when I call them ambiguous I'm am indeed
talking about the case where you recognize them as being ambiguous, I
would take it as a personal favour if you would read the suggestion I
made and post your judgement in response to that same post (not this one).
More information about the samba-technical