[geos-devel] Unclear objects lifetime and ownership issues in Monotone Chain components

[strk]

On Wed, Sep 24, 2008 at 10:50:59PM +0200, Mateusz Loskot wrote:
> strk wrote:

> > The most effective way was usually kind of "pools".
> > Basically, you use containers of things to delete at the
> > end of the operations (every operation is a class anyway).
> > It usually did fine.
> 
> Generally, true but number of issues come to my mind:
> Those pools are not documented.

Right. But you'll find a few around...

> There is no way to learn where are start and end points of operations and
> algorithms, unless you read the code from start to the end.

There kind of is, if you assume operations are driven by operation
classes...

> Memory and time boundaries of operations are not documented.

True, you may very well end out of memory before getting 
a chance to dispose all.

> So, tracking objects ownership and flows is very hard.
> How do we know all users of given pool in given time (without reading the
> whole codebase).

Which is another well known problem: there are too many interfaces
exposed. My solution was: warranty void if you don't use the C-API.

So you'd basically follow the flow from C-API on.
The wider your pools, the more issues with memory and time boundaries.
The closer-to-low-level-op you pool the better.

--strk;