Re: Ada.Containers and concurrent modification exception.

[Simon Wright <simon@pushface.org>]

rakusu_klein@fastmail.jp writes:

> Why Ada's Cursors does not provide the
> ConcurrentModificationException, as Java Collections do, or some
> variant of it? Consider the following:
>
> with Ada.Containers.Indefinite_Ordered_Maps;
> ...
>    The_Map : Map;
> ...
> declare
>    I : Cursor := First (The_Map);
>    J : Cursor := First (The_Map);
> begin
>    --  Now Cursors are synchronized with each other and with a container.
>    Delete (The_Map, I);
>    --  It's O'k. But now J lost a sync and points to a dead Node.
>    Next (J);
>    --  What should I expected here,
>    --  any well defined exception or
>    --  raising a zombie?
> end;

What actually happens in this case (GNAT CE 2018) is that the program
enters an endless loop looking at (what it thinks is) a node with both
left and right pointers designating itself.

The ARM goes to a lot of trouble to prevent "tampering with cursors",
but that's mainly to do with detecting altering the structure of a
container while iterating over it, and the code you show isn't really
covered by that. So I'm not sure if it isn't 'just' erroneous[1].

It would be a good thing if the error was detected. Perhaps submit a bug
report to AdaCore?

[1] http://www.adaic.org/resources/add_content/docs/95style/html/sec_5/5-9.html