Loading ...
Sorry, an error occurred while loading the content.

Re: [jslint] Re: Enhancement Request: Identifier length restriction

Expand Messages
  • Cheney, Edward A SSG RES USAR USARC
    I do not see the value in a minimal length requirement for identifiers even if that requirement is optional. What problems does this actually prevent? JSLint
    Message 1 of 11 , Jul 5, 2010
    • 0 Attachment
      I do not see the value in a minimal length requirement for identifiers even if that requirement is optional. What problems does this actually prevent? JSLint already catches undeclared variables and so forth regardless of the length of such. Really, what this looks like to me an abstract arbitrary name replaced by a longer name that is just as much abstract and arbitrary. In my opinion this fix merely sounds like an attempt to make code for reader friendly, which is a false expectation from such a requirement as expanding length by no means implies the longer length identifier is any more relevant a name.

      Austin Cheney
      http://prettydiff.com/
    • abyssoft@ymail.com
      Austin, Your reply looked rather vitriolic to my eyes. If it is optional why such disdain? For me I have found that with short identifiers in a long procedure
      Message 2 of 11 , Jul 5, 2010
      • 0 Attachment
        Austin,

        Your reply looked rather vitriolic to my eyes. If it is optional why such disdain?

        For me I have found that with short identifiers in a long procedure I am quite prone to accidental reuse of them in a manner that introduces bugs. By forcing myself to use longer identifiers I can also be more assured that I have not left in debug code, as that is the only time I would wish to use short identifiers. Under the use of these conditions I would find it of great benefit.

        As for abstract arbitrary names length, and casing can help. Personally I always attempt to make it as clear as possible the purpose of the identifier in it name. On the rare occasion this proves difficult I make sure to add documentation.

        For me personally it is exceedingly annoying to encounter code that has nothing but 1 to 2 char identifiers and no or poor documentation. I always wish to leave code in a better designed and documented state then when I first encountered it so that those who come after me, including myself at a later point, have an easier time modifying/correcting/adding.

        --- In jslint_com@yahoogroups.com, "Cheney, Edward A SSG RES USAR USARC" <austin.cheney@...> wrote:
        >
        > I do not see the value in a minimal length requirement for identifiers even if that requirement is optional. What problems does this actually prevent? JSLint already catches undeclared variables and so forth regardless of the length of such. Really, what this looks like to me an abstract arbitrary name replaced by a longer name that is just as much abstract and arbitrary. In my opinion this fix merely sounds like an attempt to make code for reader friendly, which is a false expectation from such a requirement as expanding length by no means implies the longer length identifier is any more relevant a name.
        >
        > Austin Cheney
        > http://prettydiff.com/
        >
      • Marc Draco
        Austin may be vitriolic, but he has a point. Code Complete by Steve McConnell is my bible and it has some interesting things to say on single-character
        Message 3 of 11 , Jul 6, 2010
        • 0 Attachment
          Austin may be vitriolic, but he has a point.

          Code Complete by Steve McConnell is my bible and it has some interesting
          things to say on single-character identifiers.

          Personally, I only ever use single-character vars in short loops. For a
          single iterator, I tend to use "n" (which betrays my beginnings in an
          early BASIC); and I would generally eschew i, j, k for nested loops
          unless there's a very good reason - i and j are so easy to confuse due
          to "confirmation bias".

          x & y (and z) are a special case also - these are handy for short loop
          iterations when co-ordinate systems are in use.

          Where speed isn't paramount, I've found a short object can be very
          helpful. Using X and Y as an example, let's say I need two versions of
          each, I might create two literal, local objects like this:

          var old = {};
          var cur = {};

          and initialise some variables

          old.x = 0;
          old.y = 0;
          cur.x = old.x +10;
          cur.y = old.x +10;

          Any operations could conceivably use "with" on these but that has
          problems all of its own. Personally I never use it.

          This is far clearer than using very long names, is very easy to type and
          less error prone because we're grouping related variables. Of course,
          your mileage may vary.
        • Michael Mikowski
          Sure James. Cut and paste from our standards document: All arguments to a function have a master prefix, arg_ . Prefixes for compound variable types prefix
          Message 4 of 11 , Jul 6, 2010
          • 0 Attachment
            Sure James. Cut and paste from our standards document:

            All arguments to a function have a master prefix, 'arg_'.

            Prefixes for compound variable types

            prefix
            variable type (vtype)
            obj_ full object with methods
            hash_ object as associative array
            ary_ array
            elem[type]_ dom element object; type is optional

            $
            jquery object


            Prefixes for scalar variable type
            prefix
            scalar type (stype)
            count_ Integer counter
            idx_ Integer loop or array index
            idint_ Integer interval id
            idto_ Integer timout id
            int_
            Integer, general

            ms_ Integer millisecond
            num_
            floating point number, general

            px_ Integer pixel units
            str_ string, general

            sw_ boolean switch (true or false)
            Examples:

            A single scalar might be px_screen_x.

            If the variable is compound then the prefixes should be strung together, e.g. ary_px_screen or hash_px_screen.

            var fnRenderScreen = function ( arg_hash_px_screen ){...};

            If a function takes multiple named arguments, we use a single input hash called arg_hash_spec, e.g:

            var fnRenderScreen = function ( arg_hash_spec ){...};

            called like so:

            ret = fnRenderScreen({ sw_overlay : true, px_left : 10, px_top: 250, px_width: 150, px_height: 150 });

            Hope that helps :)

            Cheers, Mike




            ________________________________
            From: James Friedman <james.william.friedman@...>
            To: z_mikowski@...
            Sent: Mon, July 5, 2010 3:40:50 AM
            Subject: Re: [jslint] Re: Enhancement Request: Identifier length restriction

            Hello Mike,


            I am interested in the prefixes that you use. I find the use of
            prefixes to be invaluable in keeping track of what-is-what - But then
            I do often suffer from "Teflon of the brain - nothing sticks..."


            Thank you,

            - James

            On Mon, Jul 5, 2010 at 12:53 AM, Michael Mikowski <z_mikowski@...> wrote:
            >
            >
            >
            > Hi All:
            >
            > On a bit of a tangent, there are no sigils in javascript, so therefore we use RPN to indicate data types. So my implementation would look something like this:
            >
            > var
            > ary_foo = [ 'bing','bang','boom'],
            > idx_foo
            > ;
            >
            > for ( idx_foo = 0; idx_foo < ary_foo.length; idx_foo++ ){ }
            >
            > We have a comprehensive list of prefixes we currently use, if anyone is interested.
            >
            > Feedback and suggestions are welcome :)
            >
            > Cheers, Mike
            >
            > ________________________________
            > From: "abyssoft@..." <abyssoft@...>
            > To: jslint_com@yahoogroups.com
            > Sent: Sun, July 4, 2010 3:11:36 PM
            > Subject: [jslint] Re: Enhancement Request: Identifier length restriction
            >
            > That is why I would like it as a switch/option not a requirement.
            >
            > I would write
            >
            > for (var i = 0; i < foo.length; i += 1) {}
            >
            > as
            >
            > var fooIndex;
            > for (fooIndex = foo.length -1 ; fooIndex + 1 ; fooIndex -= 1) {
            > //Loop Content even if only a one liner.
            > }
            >
            > I never declare vars in loops, or in the setup of a loop.
            >
            > --- In jslint_com@yahoogroups.com, "pauanyu" <pcxunlimited@...> wrote:
            > >
            > >
            > >
            > > What about this (very) common construct:
            > >
            > > for (var i = 0; i < foo.length; i += 1) {}
            > >
            > > I agree that very short identifiers are generally a bad idea, but I also think there are a couple exceptions (like the above).
            > >
            > > --- In jslint_com@yahoogroups.com, "abyssoft@" <abyssoft@> wrote:
            > > >
            > > > Reasoning
            > > >
            > > > Identifiers of only 1 or 2 Characters can lead to errors in thought and poor readability of code. I have a preference for identifiers that consist of 3 or more characters and they are rarely that short.
            > > >
            > > > An alternate enhancement is for support of single and double letter identifiers when listed in /*global*/.
            > > >
            > >
            >
            > [Non-text portions of this message have been removed]
            >
            >


            --
            --------------------------
            Television and the media culture have replaced thoughtfulness with
            entertainment. We are being converted from a critically minded
            inquisitive population enjoying informed discourse to a nation of
            passive retards.


            [Non-text portions of this message have been removed]
          • Cheney, Edward A SSG RES USAR USARC
            ... Beautify the code using a JSLint white space compatible algorithm. Then perform the following replacements: For functions: - Manually change the name of
            Message 5 of 11 , Jul 6, 2010
            • 0 Attachment
              > For me personally it is exceedingly annoying to encounter code that
              > has nothing but 1 to 2 char identifiers and no or poor documentation.

              Beautify the code using a JSLint white space compatible algorithm. Then
              perform the following replacements:

              For functions:
              - Manually change the name of the function where it is declared
              - Find and replace on the following after ensure that each of the
              following is not in quotes:
              + space + variable name + left paren
              ex: " a("
              + space + variable name + comma
              ex: " a,"
              + space + variable name + right paren
              ex: " a)"
              + space + variable name + right square brace
              ex: " a]"
              + space + variable name + right curly brace
              ex: " a}"

              For arrays:
              - Same as above except find and replace on a left square brace
              instead of opening paren

              For object literals
              - Same as above except find and replace on a left curly brace instead
              of opening paren

              For value literals (strings and numbers)
              - Same as above except find and replace on a space, instead of an
              opening paren

              You could write a JS application to do all of this for you. This would
              be the best bet because you are not going to influence how other people
              write code their own code. The only code that you can effectively
              influence is the code under your own responsibility.

              Austin Cheney
              http://prettydiff.com/
            Your message has been successfully submitted and would be delivered to recipients shortly.