REQUEST: better empty catch in better exception handling

[Jeremy Manson]

Do people do empty catch blocks by mistake?  They always think it is
right, even if it isn't.  I just imagine people sprinkling this
annotation everywhere and not getting anything out of it.

Jeremy

On Tue, Mar 24, 2009 at 6:22 PM,  <rssh at gradsoft.com.ua> wrote:
>
>  Yet one problem with extension handler: rare we need catch exception but do
> nothibg in catch statement. (i. e. do not log)
> Example is code like next snipset:
>
> try {
>  something=searchFirst(condition);
> }catch(NotFoundException ex){
>  // do nothing. search better
> }
>
> if (something!=null) {
>  try {
>  something=searchSecond(condition);
>  }catch(NotFoundException ex){
>  // do nothing. search better
>  }
> }
>
> >From other side, we prefer do not have in codebase code with empty
> exception handling and use special tools to detect empty catch braclets
> and warn about ones.
>  For now,   we use ';' to mark, that this empty catch is really empty
> catch and not error or unfinished refactoring. I.e.:
>
> if (something!=null) {
>  try {
>  something=searchSecond(condition);
>  }catch(NotFoundException ex){
>  // do nothing. search better
>  ;  // ';' tell javachecker do not complain about empty catch.
>  }
> }
>
>
> But this is ugly.
> Of course, is simple use annotation
> @SuppressWarning("empty-catch")  but javac compiler does not warn about
> empty catch-s.
>
> So, question to community: is it possible to extend exception handling
> proposal by adding optional conventions
>  - warn about empty catch-s
>  - specify rare cases, when empty catch is really needed. ?
>
> For example, this can be @EmptyCatch annotation, which mark catch block
> empty. i.e.
> try {
>
> }catch(@EmptyCatch  NotFound | FoundNotAsILike ex ){
> }
> and effective do nothing.
>
>
>
>
>
>