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

Re: [jslint] option.stupid

Expand Messages
  • Martin Cooper
    On Fri, Mar 30, 2012 at 11:08 AM, douglascrockford ... There are numerous valid and reasonable use cases for synchronous code in Node.js. To declare the use of
    Message 1 of 11 , Mar 30, 2012
    • 0 Attachment
      On Fri, Mar 30, 2012 at 11:08 AM, douglascrockford
      <douglas@...> wrote:
      > JSLint now warns when properties contain the substring 'Sync'. The use of that substring in Nodejs is to identify methods that can cause blockage. Such methods should never be used.
      >

      There are numerous valid and reasonable use cases for synchronous code
      in Node.js. To declare the use of sync functions to be universally
      "stupid" suggests a distinct lack of consideration for the breadth of
      use cases that Node.js is being used to address.

      Two very simple and very common examples of where synchronous
      functions are frequently used in Node.js code are command line tools,
      for which async often makes little sense and only complicates the
      code, and loading server config at start-up time, for similar reasons.

      In both of those cases, you *could* use async code. But when there is
      no good reason for the code to be async, and the sync code would be
      simpler and easier to write, debug, read and maintain, it is
      absolutely not "stupid" to make the decision to use the synchronous
      functions.

      To tell people that they need to "tolerate stupidity" in order to take
      the most appropriate approach for their use cases is inappropriate at
      best.

      --
      Martin Cooper


      > These useful warnings can be suppressed by using the new Tolerate stupidity option.
      >
      >
      >
      >
      >
      > ------------------------------------
      >
      > Yahoo! Groups Links
      >
      >
      >
    • Mark Volkmann
      I agree with Martin. Why not fix this by changing the name of the option to sync and have it default to false? ... R. Mark Volkmann Object Computing, Inc. On
      Message 2 of 11 , Mar 30, 2012
      • 0 Attachment
        I agree with Martin. Why not fix this by changing the name of the option to
        "sync" and have it default to false?

        ---
        R. Mark Volkmann
        Object Computing, Inc.

        On Mar 30, 2012, at 7:55 PM, Martin Cooper <mfncooper@...> wrote:



        On Fri, Mar 30, 2012 at 11:08 AM, douglascrockford
        <douglas@...> wrote:
        > JSLint now warns when properties contain the substring 'Sync'. The use of
        that substring in Nodejs is to identify methods that can cause blockage.
        Such methods should never be used.
        >

        There are numerous valid and reasonable use cases for synchronous code
        in Node.js. To declare the use of sync functions to be universally
        "stupid" suggests a distinct lack of consideration for the breadth of
        use cases that Node.js is being used to address.

        Two very simple and very common examples of where synchronous
        functions are frequently used in Node.js code are command line tools,
        for which async often makes little sense and only complicates the
        code, and loading server config at start-up time, for similar reasons.

        In both of those cases, you *could* use async code. But when there is
        no good reason for the code to be async, and the sync code would be
        simpler and easier to write, debug, read and maintain, it is
        absolutely not "stupid" to make the decision to use the synchronous
        functions.

        To tell people that they need to "tolerate stupidity" in order to take
        the most appropriate approach for their use cases is inappropriate at
        best.

        --
        Martin Cooper

        > These useful warnings can be suppressed by using the new Tolerate
        stupidity option.
        >
        >
        >
        >
        >
        > ------------------------------------
        >
        > Yahoo! Groups Links
        >
        >
        >



        [Non-text portions of this message have been removed]
      • Andy Dawson
        Alternatively, all the other options could be renamed to follow the same convention so that all uses can consistently need to lookup what options mean instead
        Message 3 of 11 , Mar 31, 2012
        • 0 Attachment
          Alternatively, all the other options could be renamed to follow the same
          convention so that all uses can consistently need to lookup what options
          mean instead of having some hint from the option name.

          as such, lameness, awfulness, javaitis etc. can all become part of the
          jslint vocabulary

          AD
          On Mar 31, 2012 4:08 AM, "Mark Volkmann" <r.mark.volkmann@...> wrote:

          > I agree with Martin. Why not fix this by changing the name of the option to
          > "sync" and have it default to false?
          >
          > ---
          > R. Mark Volkmann
          > Object Computing, Inc.
          >
          > On Mar 30, 2012, at 7:55 PM, Martin Cooper <mfncooper@...> wrote:
          >
          >
          >
          > On Fri, Mar 30, 2012 at 11:08 AM, douglascrockford
          > <douglas@...> wrote:
          > > JSLint now warns when properties contain the substring 'Sync'. The use of
          > that substring in Nodejs is to identify methods that can cause blockage.
          > Such methods should never be used.
          > >
          >
          > There are numerous valid and reasonable use cases for synchronous code
          > in Node.js. To declare the use of sync functions to be universally
          > "stupid" suggests a distinct lack of consideration for the breadth of
          > use cases that Node.js is being used to address.
          >
          > Two very simple and very common examples of where synchronous
          > functions are frequently used in Node.js code are command line tools,
          > for which async often makes little sense and only complicates the
          > code, and loading server config at start-up time, for similar reasons.
          >
          > In both of those cases, you *could* use async code. But when there is
          > no good reason for the code to be async, and the sync code would be
          > simpler and easier to write, debug, read and maintain, it is
          > absolutely not "stupid" to make the decision to use the synchronous
          > functions.
          >
          > To tell people that they need to "tolerate stupidity" in order to take
          > the most appropriate approach for their use cases is inappropriate at
          > best.
          >
          > --
          > Martin Cooper
          >
          > > These useful warnings can be suppressed by using the new Tolerate
          > stupidity option.
          > >
          > >
          > >
          > >
          > >
          > > ------------------------------------
          > >
          > > Yahoo! Groups Links
          > >
          > >
          > >
          >
          >
          >
          > [Non-text portions of this message have been removed]
          >
          >
          >
          > ------------------------------------
          >
          > Yahoo! Groups Links
          >
          >
          >
          >


          [Non-text portions of this message have been removed]
        • Morgaut Alexandre Louis Marc
          I find this option name fun and maybe It could be a right name in NodeJS context. That said, I also think that there might be some valid use of Sync methods
          Message 4 of 11 , Apr 2, 2012
          • 0 Attachment
            I find this option name fun and maybe It could be a right name in NodeJS context.

            That said, I also think that there might be some valid use of "Sync" methods in JavaScript. I'm thinking about the Web Workers, and the synchronous SSJS platforms.

            I know you may not have much consideration for HTML5 (mainly because security should have been addressed in priority) but still...

            In HTML5, many APIs propose Sync and Async modes:
            - XMLHttpRequest 1 & 2: http://www.w3.org/TR/XMLHttpRequest/ - http://www.w3.org/TR/XMLHttpRequest2/
            -> it would not be impacted by this option as is because "Sync" isn't part of property names in the API
            - File API: http://www.w3.org/TR/FileAPI/
            - File System API: http://www.w3.org/TR/file-system-api/
            - File Writer API: http://www.w3.org/TR/file-writer-api/
            - Indexed Database API: http://www.w3.org/TR/IndexedDB
            - Web SQL Database (inactive): http://www.w3.org/TR/webdatabase/

            This didn't make much sense before because it was blocking the User Interface, but now, we have the Web Workers allowing to take advantage of multi-threading (or at least multi processing). It then looks to me that using synchronous APIs in Workers should be perfectly acceptable, even more when we know that the importScript() method available in this context works synchronously.


            Another point is about other Server-Side JavaScript implementations
            Most of them have synchronous APIs and take advantage of multi-threading
            I'm not aware of all the details of all of them, most should not get errors from this option as they don't have "Sync" in their method names. In this case only the option name "stupid" is a bit hard (should our feelings being hurt when we don't choose NodeJS?)
            To end my purpose, I would also mention Wakanda, which runs JS code on server synchronously by default. It support natively some HTML5 and NodeJS APIs to:
            - start event loops like contexts when it make sense,
            - be compatible with some existing libraries,
            - and of course enhance learning curve.
            On this platform, not only most of the JavaScript works synchronously, but it would also get the warning while using the synchronous versions of HTML5 APIs.
            Maybe a warning is ok, because such code could not be reused in any context (main user interface context, or Node.js context), but then, isn't the "stupid" option name a bit too strong?
          • benquarmby
            Doug, I don t have a problem with this option, but I would like to suggest a stricter implementation of it s behavior. Instead of broadly catching the
            Message 5 of 11 , Jun 20 5:46 PM
            • 0 Attachment
              Doug,

              I don't have a problem with this option, but I would like to suggest a stricter implementation of it's behavior.

              Instead of broadly catching the substring "Sync", limit it to catch functions that *end* with "Sync", since all blocking Node methods appear use this convention.

              Matching the end will also stop it catching false positives. For example, the completion event handler for an asynchronous synchronize action, i.e. "onGroupSyncComplete".

              Hope you agree.

              --- In jslint_com@yahoogroups.com, "douglascrockford" <douglas@...> wrote:
              >
              > JSLint now warns when properties contain the substring 'Sync'. The use of that substring in Nodejs is to identify methods that can cause blockage. Such methods should never be used.
              >
              > These useful warnings can be suppressed by using the new Tolerate stupidity option.
              >
            Your message has been successfully submitted and would be delivered to recipients shortly.