[Laszlo-dev] for review, Change 20071028-hqm-6, put XML parser back into server data proxy pipeline, to do charset transcoding

[Mamye Kratt]

It cannot be reproduced everytime. It started when we moved to the vm server (margrathea). I've been tracking it, but cannot make it happen. The build system currently take 2 plus hours to run. The vm machine also fails to build when multiple users are working on it (not sure how many users, tracking this also). I think this is related to the settings in the limits file.

m
  ----- Original Message ----- 
  From: Henry Minsky 
  To: Mamye Kratt 
  Cc: Laszlo-dev bug reporting ; Max Carlson 
  Sent: Wednesday, October 31, 2007 12:27 PM
  Subject: Re: [Laszlo-dev] for review, Change 20071028-hqm-6,put XML parser back into server data proxy pipeline,to do charset transcoding


  Hmm, well that's not really what I'm talking about, but we
  ought to look into that. 

   Where would I be able to see what
  "svn-get-source" is doing? Is that a shell function, or ant target, or something? 




  On 10/31/07, Mamye Kratt <mamye at laszlosystems.com> wrote:
    The svn error is below:

    On Oct 22, 2007, at 3:04 PM, Mamye Kratt wrote:


      svn-get-source:
          [exec] svn: REPORT request failed on '/!svn/vcc/default'
          [exec] svn: REPORT of '/!svn/vcc/default': Could not read chunk size: Connection was closed by server. (http://svn.openlaszlo.org)

       

      ----- Original Message ----- 
      From: Henry Minsky 
      To: Max Carlson ; Laszlo-dev bug reporting 
      Sent: Wednesday, October 31, 2007 8:07 AM
      Subject: Re: [Laszlo-dev] for review, Change 20071028-hqm-6,put XML parser back into server data proxy pipeline,to do charset transcoding


      Note, that this code does not know the content length before it starts writing the output to
      the client, since it's being streamed.  So it doesn't set the content length explicitly, like it used to.

      My understanding of Tomcat's behavior is that if the content is small enough to fit in some tomcat output 
      buffer size , it will buffer it and then send it all with a content length header. If it is larger than that
      then Tomcat sends it in HTTP chunked encoding. I still have to verify this with a very large file. 



      On 10/28/07, Henry Minsky <henry.minsky at gmail.com> wrote: 
        Change 20071028-hqm-6 by hqm at IBM-2E06404CB67 on 2007-10-28 20:42:12 EDT
            in /cygdrive/c/users/hqm/openlaszlo/trunk
            for http://svn.openlaszlo.org/openlaszlo/trunk 

        Summary: put XML parser back into server data proxy pipeline, to do
        charset transcoding

        New Features:

        Bugs Fixed: LPP-4924

        Technical Reviewer: max
        QA Reviewer: pablo
        Doc Reviewer: 

        Details:

        The server data proxy now uses the XMLPULL parser to parse
        the data from the backend, in order to use Java to force a translation
        into UTF-8
        coding if needed.

        This change also uses a worker Thread to read from the backend, while 
        simultaneously
        pipelining the data back throug the XML PULL parser to the client.
        This should improve
        response time and also removes a potential memory overflow and DOS
        attack on the server.

        Tests:

        test/lfc/data/alldata.lzx
        amazon
        calendar


        Files:
        D      WEB-INF/lib/xpp3-1.1.3.4d_b4.jar
        A      WEB-INF/lib/xpp3-1.1.4c.jar
        M      WEB-INF/lps/lfc/kernel/swf/LzLoadQueue.as 
        A      WEB-INF/lps/server/src/org/openlaszlo/data/HttpData.java 
        M      WEB-INF/lps/server/src/org/openlaszlo/data/XMLGrabber.java
        M      WEB-INF/lps/server/src/org/openlaszlo/data/HTTPDataSource.java
        M      WEB-INF/lps/server/src/org/openlaszlo/data/DataSource.java



        --
        Henry Minsky
        Software Architect
        hminsky at laszlosystems.com




      -- 
      Henry Minsky
      Software Architect
      hminsky at laszlosystems.com





  -- 
  Henry Minsky
  Software Architect
  hminsky at laszlosystems.com

-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://www.openlaszlo.org/pipermail/laszlo-dev/attachments/20071031/467373d7/attachment-0001.html