[crac] RFR: Fix ordering of invocation on Resources [v10]

[Anton Kozlov]

On Wed, 10 May 2023 13:48:08 GMT, Radim Vansa <duke at openjdk.org> wrote:

>> It seems we should have allowed the message in the CheckpointException, to avoid this problem. The only place where CheckpointException with the message is called is recursive checkpoint detection from https://github.com/openjdk/crac/pull/6. 
>> 
>> OK for this PR.
>
> That's not entirely true; CheckpointException is a public API and Resources can use it. In fact I find it quite natural to use for reporting any 'generic' error during checkpoint; if we want to limit usage to internal we need to make the constructors package-private constructors.

Hiding all constructors makes sense. I think it will be helpful to have CheckpointException, CheckpointResourceExceptions and descedants something formal, something that can be programmatically queried for the problem. I propose to investigate the possiblity in #64 which touches related pieces.

-------------

PR Review Comment: https://git.openjdk.org/crac/pull/60#discussion_r1190070472