Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]

Groups > comp.lang.javascript > #38393

Exception handling (was: check for obj nested prop)

Show key headers only | View raw

Ben Bacarisse wrote:

> Andrew Poulos <ap_prog@hotmail.com> writes:
>> let obj = {a: 1};
>> if (a.b.c.d) // throws error
> 
> Did you mean if (obj.a.b.c.d) ?

Probably yes.

>> so I end up doing
>>
>> if (a && a.b && a.b.c && a.b.c.d)
> <cut>
>> Is there a way to get the check value of an object's nested property
>> without first having to test for all the ascendant properties?
> 
>   if (try {obj.a.b.c.d} catch {})
> 
> might do what you want.

By contrast, the following is syntactically correct:

  try {
    if (obj.a.b.c.d)
    {
      // …
    }
  } catch {}

However, there are several caveats:

1. The omission of the exception parameter requires an implementation of
   ECMAScript Edition 10 (ECMAScript 2019).  This is the most recent
   Edition of the standard and requires the most recent version of a runtime
   environment (e.g. a Web browser) to have a chance to be implemented.

   Compare

     <http://ecma-international.org/ecma-262/9.0/#prod-Catch>

   with

     <http://ecma-international.org/ecma-262/10.0/#prod-Catch>

2. Using an empty try-catch block can hide problems in code that are not
   syntax errors, and is strongly recommended against (as in “SHOULD NOT”). 
   For example, the code above hides a problem of an undeclared identifier
   “obj” even in strict mode, which would throw a ReferenceError exception.

   Therefore, if try-catch is used, in almost all cases exception handling
   should be specific.

   Unfortunately, ECMAScript did not standardize a feature of Mozilla
   JavaScript where it was possible to write

     catch (e if e instanceof ExceptionType) {
       // …
     }

   where ExceptionType would be the exception type, e.g. TypeError,
   and as of Mozilla JavaScript 68 (Firefox 68) it is removed from
   the implementation.

   However,

     try {
     }
     catch (e) {
       if (e instanceof ExceptionType) {
         // …
       }
     }

    is functionally equivalent and strictly standards-compliant
    (specified explicitly in the standard).

    Finally, exceptions that are not handled should be re-thrown so as to
    not hide more serious problems:
  
     catch (e) {
       if (e instanceof ExceptionType) {
         // …
       }
       else if (e instanceof …) {
         // …
       }
       else throw e;
     }

3. With that said, it is actually not easy to tell exception conditions 
   apart.  For example, if a property-read access results in the “undefined”
   value because an object does not have a property with the specified name,
   then the subsequent lookup of a “nested” property will throw a TypeError
   exception because “undefined has no properties” (it is a primitive
   value that cannot be converted to an object).

   But a TypeError exception is being thrown by several other operations as
   well (see the standard’s prose), so the recommended *alternative* code

     try {
       if (obj.a.b.c.d)
       {
         // …
       }
     } catch (e) {
       if (e instanceof TypeError) {
         // …
       }
       else throw e;
     }

   is *unable* to tell those conditions apart, and there is none more
   specific exception type, e.g. a “KeyError“ exception (as it is available
   for example in Python).

Surprising to me, the Airbnb JavaScript Style Guide¹ makes no explicit 
recommendation or requirement for exception handling.  However, they refer 
to a jsPerf test case for “Try/Catch Cost In a Loop”:

<https://jsperf.com/try-catch-in-loop-cost/12>

Also, you should be aware that while ESLint² supports defining several rules 
for “throw” and “catch”³, none of them can protect against the problems with 
exception handling that I discussed above.

______
¹ <https://github.com/airbnb/javascript>
² <https://eslint.org/>
³ <https://eslint.org/docs/rules/>
 
-- 
PointedEars
FAQ: <http://PointedEars.de/faq> | <http://PointedEars.de/es-matrix>
<https://github.com/PointedEars> | <http://PointedEars.de/wsvn/>
Twitter: @PointedEars2 | Please do not cc me./Bitte keine Kopien per E-Mail.

Back to comp.lang.javascript | Previous | Next — Previous in thread | Next in thread | Find similar | Unroll thread

Thread

check for obj nested prop Andrew Poulos <ap_prog@hotmail.com> - 2020-05-08 15:37 +1000
  Re: check for obj nested prop Ben Bacarisse <ben.usenet@bsb.me.uk> - 2020-05-08 10:52 +0100
    Re: check for obj nested prop Thomas 'PointedEars' Lahn <PointedEars@web.de> - 2020-05-09 00:26 +0200
      Re: check for obj nested prop Ben Bacarisse <ben.usenet@bsb.me.uk> - 2020-05-09 03:09 +0100
        Re: check for obj nested prop Thomas 'PointedEars' Lahn <PointedEars@web.de> - 2020-05-12 09:28 +0200
          Re: check for obj nested prop Julio Di Egidio <julio@diegidio.name> - 2020-05-12 04:01 -0700
    Exception handling (was: check for obj nested prop) Thomas 'PointedEars' Lahn <PointedEars@web.de> - 2020-05-09 03:46 +0200
  Re: check for obj nested prop JJ <jj4public@vfemail.net> - 2020-05-08 17:18 +0700
  Re: check for obj nested prop Julio Di Egidio <julio@diegidio.name> - 2020-05-08 03:53 -0700
  Re: check for obj nested prop James Kirk <noneya.invalid@gmail.com> - 2020-05-08 11:00 -0400
  Re: check for obj nested prop Thomas 'PointedEars' Lahn <PointedEars@web.de> - 2020-05-09 00:19 +0200
    Re: check for obj nested prop Julio Di Egidio <julio@diegidio.name> - 2020-05-12 04:04 -0700
      Re: check for obj nested prop luserdroog <luser.droog@gmail.com> - 2020-05-12 07:59 -0700
  Re: check for obj nested prop "Michael Haufe (TNO)" <tno@thenewobjective.com> - 2020-05-12 14:17 -0700
    Re: check for obj nested prop Thomas 'PointedEars' Lahn <PointedEars@web.de> - 2020-05-13 02:46 +0200
    Re: check for obj nested prop Andrew Poulos <ap_prog@hotmail.com> - 2020-05-13 13:08 +1000
      Re: check for obj nested prop "Michael Haufe (TNO)" <tno@thenewobjective.com> - 2020-05-13 12:30 -0700

csiph-web