                                                                                                                                                                                                                                                                
Received-SPF: pass (google.com: domain of alaku...@yandex.ru designates 77.88.61.38 as permitted sender) client-ip=77.88.61.38;
MIME-Version: 1.0
In-Reply-To: <CAJJvDZEM1ZKJAS8Bj2X4j3SdW7oXmGg=r+hTPTi0ea0b-7t...@mail.gmail.com>
References: <24544524-5acc-4e09-a9e3-7dd661867d87@googlegroups.com>
	<CAJJvDZHB+8CcRLHTC+RbS=JB=XNHQTAnTdJwqzissBScGJ6...@mail.gmail.com>
	<CAJJvDZFacRxX8Uzv2eGj8bcZ7KsEzmfRvUdKZsbuyHbEAMU...@mail.gmail.com>
	<36c9b20a-533b-42ec-838b-414f8e0e45e8@googlegroups.com>
	<CAJJvDZEM1ZKJAS8Bj2X4j3SdW7oXmGg=r+hTPTi0ea0b-7t...@mail.gmail.com>
Date: Tue, 24 Jul 2012 12:55:21 +0400
Message-ID: <CAJsdZpcMzqKvw90oz0UL7pSS0XnE9DbxaL4Qd70dJNW13mg...@mail.gmail.com>
Subject: Re: [closure-compiler-discuss] False "Missing return statement" warning
From: Alexey Lakutin <alaku...@yandex.ru>
To: closure-compiler-discuss@googlegroups.com
Content-Type: text/plain; charset=ISO-8859-1

Well, I understand this, but not sure about compiler.
Actually, I do not need to throw and return value at the same time.
And I do not want to catch exception here.
Guess, I understand Compiler thoughts about my code:
it is how "finally" block processed.
Probably Compiler knows that "finally" is the last block of function
(in execution order), but
it does not take into account that it is NOT the block that should
return a value and erroneously expect it,
though value is returned by "previous" execution block, which is the
last statement of normal execution flow.
Not sure is it bug of compiler or not.
Guess it should check "normal" and "catch" execution paths for return
value, but omit "finally" block.

2012/7/24 Chris Price <pric...@gmail.com>:
> The difference is that finally doesn't swallow the exception so the
> function never returns any value, instead it throws the exception [1].
> As far as I'm aware there isn't any way to propagate the exception at
> the same time as returning a value, it just doesn't make sense. You
> could instead catch the exception and then return false [2].
>
> [1] http://jsfiddle.net/tVdpb/
> [2] http://jsfiddle.net/tVdpb/1/
>
> On Tue, Jul 24, 2012 at 8:58 AM, LioLick <alexey.laku...@gmail.com> wrote:
>> Thanks Chris,
>> it is just a minimal sample code that can reproduce such a problem.
>> I'm trying to compile large amount of existing code with Closure Compiler.
>> I've discovered that moving "return true;" into finally block resolves the
>> problem, but still want to know - what's the difference from Closure
>> Compiler point:
>> code flow is the same in both cases.
>> @throws tag does not satisfy compiler.
>> Common use case is:
>> try
>> {
>>   // some dangerous operation
>> }
>> finally
>> {
>>   // restore initial state in both cases: success or failure
>> }
>>
>> // postprocess result on success or pass exception up
>> return result;