我有以下代码,我需要扫描抛出的异常。如果它满足特定条件,我将忽略该异常。否则我重新扔。抛出的异常是一个检查异常,这意味着重新抛出是困难的部分。包括我必须抓住它,并且操作发生在一个被覆盖的方法中,该方法的基类方法没有使用throws clause
. 卡在重新编程整个庞大的库或将异常包装为RuntimeException
(这不是一个选项,因为我们将包装的检查异常,在其传播过程中的某个地方(派生类)预期会被一个 catch 子句 - 它的某种信号),我希望就这样的实现得到一些建议。或者也许只是实施。
/*
* Translated from c++11_7.3.19 AIEDs schemes >${ref[213]:`through}.
* guarantees learning, re-learning and 'un-learning'. programmed intelligence is not
* altered whatsover and is made as root of bias [x:~y]. fetch cycle is omitted.
*/
@Override
public void intelligenceSync()// No throws clause here
{
try {
super.intelligenceSync();
// the throwing culprit.
global_contribution.acceptOrReformState(this);
generateMetaLogicReforms(this);
} catch (Throwable t_) {
if (t_ instanceof Signalx) {
Signalx sx = (Signalx) t_;
// Note how bias inreases propagation speed by ~12.21 >${ref[371]:exp2}.
applyBias(sx);
stopInvalidation(sx);
// check if x neuron is almost proved.
if (sx.neuronSP() > sx.threshold()) {
// We'll find other ways of completing our proofs.
// Note the algorithm is not so complete.
netsync(sx);
while (sx.pushDeeper()) {
sx.enhance(Metrics.COMPLETION.esteem());
sx.suspendSubPathTraversal(Suspender.IN_DREAM_RESOLVE, Handler.NULL_LOGIC);
}
generateSubLogicReforms(sx);
} else {
restore(sx);
continueInvalidation(sx);
// We rethrow.
uncheckedThrow(sx);
// exception thrown
}
} else if (t_ instanceof Signaly) {
// Reforms must be handle explicitly.otherwise RelationalAttender will complain
// .
// ... blah blah blah.
} else {
// We rethrow
uncheckedThrow(t_);
}
//
}
}