Beefy Boxes and Bandwidth Generously Provided by pair Networks
Do you know where your variables are?
 
PerlMonks  

Seeking design improvements for Perl code library remote server

by daluu (Initiate)
on Dec 25, 2012 at 04:32 UTC ( #1010245=perlquestion: print w/ replies, xml ) Need Help??
daluu has asked for the wisdom of the Perl Monks concerning the following question:

I built a custom XML-RPC server in Perl that "supposedly" serves any given Perl code module/class whose structure conforms to some compatible specifications, and the server serves it dynamically at startup via reflection/introspection. It's built out of code snippets found on this site along with a Wikipedia page, with my glue code to mash everything together. It's also modeled (in terms of design) after my server implementations for other languages (.NET, Java). For this post, I'm asking for feedback from Perl community what improvements, if any, I could make to the server. Suggestions on how to implement bug fixes for the open issues of the server, and if there is a more optimal approach to reflection for executing a Perl class method or listing out the class methods information. As the material I based it on and Perl language itself has more than one way to do things.

In terms of optimal code reflection, I'm looking for optimizing or solution where it can satisfy a majority of Perl code or modules, w/ no or minimal changes or wrappers to the Perl code being reflected in most cases.

I guess this is kind of a broad question post, but in a nutshell just wanted feedback for improvements to my server. Don't know where else is good place to seek feedback for Perl.

http://code.google.com/p/plrobotremoteserver/source/browse/#svn%2Ftrunk

http://code.google.com/p/plrobotremoteserver/wiki/UsageInfo

http://code.google.com/p/plrobotremoteserver/issues/list

and the server is built against the remote library interface spec for RobotFramework

http://code.google.com/p/robotframework/wiki/RemoteLibrary

with the reference implementation (to the spec) being in Python.

Comment on Seeking design improvements for Perl code library remote server
Re: Seeking design improvements for Perl code library remote server
by CountZero (Bishop) on Dec 25, 2012 at 20:18 UTC
    At the very least you could have made the links to the code active. Now we have to copy and paste these into the browser and I am lazy.

    Why don't you pack it all up into a proper Perl-distribution and put it on CPAN?

    CountZero

    A program should be light and agile, its subroutines connected like a string of pearls. The spirit and intent of the program should be retained throughout. There should be neither too little or too much, neither needless loops nor useless variables, neither lack of structure nor overwhelming rigidity." - The Tao of Programming, 4.1 - Geoffrey James

    My blog: Imperial Deltronics
        Please, rather than critizing my position, go ahead and review that code: download it, try to install it, debug it and let us all know. Good karma to you for your efforts.

        CountZero

        A program should be light and agile, its subroutines connected like a string of pearls. The spirit and intent of the program should be retained throughout. There should be neither too little or too much, neither needless loops nor useless variables, neither lack of structure nor overwhelming rigidity." - The Tao of Programming, 4.1 - Geoffrey James

        My blog: Imperial Deltronics

        Yes, I'm looking for code review feedback. It's hard to find for a project when you're the sole developer and it lacks a community of users using it.

        It's a remote server for Robot Framework test framework, and from what I can tell, there are potential Perl users for that, but it's relatively small compared to Java, .NET, and Python. :(

        And the main Robot Framework community developers are not Perl developers and don't have the time to help code review. :(

        I am not a server specialist, so I am not the right Monk to comment on that aspect of your code.

        One thing I am curious about is why you commented out use strict; use warnings;?

        CountZero

        A program should be light and agile, its subroutines connected like a string of pearls. The spirit and intent of the program should be retained throughout. There should be neither too little or too much, neither needless loops nor useless variables, neither lack of structure nor overwhelming rigidity." - The Tao of Programming, 4.1 - Geoffrey James

        My blog: Imperial Deltronics

        Your package doesn't match the filename, see Simple Module Tutorial , the filename should match the package name.

        Instead of comments write documentation :) Tutorials: POD in 5 minutes and maybe

        See Modern Perl and Class::Inspector AND Howto create a Moose baseclass/superclass/contract? Moose equivalent of Module::Pluggable? Favorite way to create base/super class?

        You might start with

        module-starter --author="David Luu" --email=man...mail.com --license=LGPL --module=RobotRemoteServer,RobotRemoteServer::Frontier,RobotRemoteServer,RobotRemoteServer::RPCXMLServer

        Added to MANIFEST: Changes
        Added to MANIFEST: ignore.txt
        Added to MANIFEST: lib/RobotRemoteServer.pm
        Added to MANIFEST: lib/RobotRemoteServer/Frontier.pm
        Added to MANIFEST: lib/RobotRemoteServer/RPCXMLServer.pm
        Added to MANIFEST: Makefile.PL
        Added to MANIFEST: MANIFEST
        Added to MANIFEST: README
        Added to MANIFEST: t/00-load.t
        Added to MANIFEST: t/boilerplate.t
        Added to MANIFEST: t/manifest.t
        Added to MANIFEST: t/pod-coverage.t
        Added to MANIFEST: t/pod.t
        Created starter directories and files
        

        Then maybe write something like


        package RobotRemoteServer;
        use Class::Inspector;

        sub get_keyword_names { my ($self) = @_; my $class = $self->lib; my @methods = Class::Inspector->methods( $class, 'public' ); return \@methods; } sub run_keyword { my $self = shift; my ($stdout, $stderr) = capture { ... }; } sub start_server { my $self = shift; $self->backend->start_server; }

        package RobotRemoteServer::Frontier; sub _shut_down_request { RobotRemoteServer::Frontier sub start_server { my ($self) = @_; my $svr = Frontier::Daemon->new( methods => { get_keyword_names => sub { $self->get_keyword_names(@_) }, run_keyword => sub { $self->run_keyword($_[0], eval{@{$_[1]}}) } +, stop_remote_server => sub { $srv->_shutdown('yes please'); $self->schedule_shutdown; return; }, .... } sub schedule_shutdown { my $self = shift; my $url = $self->url; if( eval { require threads; } ){ threads::async { sleep 5; require Frontier::Client; Frontier::Client->new( 'url' => $url )->call('stop_remote_s +erver'); }; } elsif( eval { require Proc::Background; } ){ Proc::Background->new( $^X, '-e', q{ use Frontier::Client; sleep 5; my $url = shift @ARGV; Frontier::Client->new( 'url' => $url )->call('stop_remote_server'); exit; }, $url, ); } } sub Frontier::Daemon::_shutdown { my( $self, $true ) = @_; ${*$self}{'_Frontier::Daemon::_shutdown'} = !!$true if $true; return ${*$self}{'_Frontier::Daemon::_shutdown'} ; } sub Frontier::Daemon::accept { ## monkeypatch package Frontier::Daemon; my $self = shift; if( $self->_shutdown ){ close $self; return; } else { return $self->SUPER::accept; } }

        To handle shutdown in similar way as https://robotframework.googlecode.com/hg/tools/remoteserver/robotremoteserver.py you could monkeypatch shutdown, but writing your own frontier deaemon ( Frontier::Daemon::Forking ) is probably the correct way to do it

        backend is the new argument, RobotRemoteServer::Frontier by default

        use RobotRemoteServer; #my $rob = RobotRemoteServer->new (qw' host localhost port 666 '); my $rob = RobotRemoteServer->new (qw' host localhost port 666 backend Frontier lib ExampleLibrary '); $rob->start_server;

        Maybe it makes more sense for backends to be named as RobotRemoteServer::Backend::Frontier ...

        But that is a start, pod beats comments, seperate backends into seperate modules, use existing modules for introspection , has-a not isa-a ...

        Future direction:
        maybe Moo/Moose or Badger for accessors/mutators...
        Maybe POE server ( ex Hopkins::Plugin::RPC ), ex?Catalyst::Plugin::Server::XMLRPC, Plack for ideas
        maybe cooperation with Daemon::Generic, AnyEvent::XMLRPC, ...

        If I were more knowledgeable, I might make better suggestions ;)

Re: Seeking design improvements for Perl code library remote server
by flexvault (Parson) on Dec 29, 2012 at 15:08 UTC

    Welcome daluu,

    First, I don't have a clue about what you are trying to do, but since you do, maybe you could explain why your approach is 'better' or 'easier' or 'friendlier' or 'whatever' for your potential users.

    One way to find the value of your code, is to profile your server with 'Devel::NYTProf'. This great tool will help you find your server's bottlenecks and help you improve your server performance. Set a realistic benchmark like ___ (your input needed) serves per minute, and then use the detailed information from 'Devel::NYTProf' to improve your own code. Look for code that can be moved outside of a loop and at the same time it will help you improve your Perl coding skills.

    Try to improve your code by 100%. If you only get 1-2% then your code is optimized already. But be honest with yourself.

    When you find something that you think could be better, but don't know how to improve the code -- that's when to ask the PM members for specific help. You'll get some good answers and a lot more help.

    Good Luck!

    "Well done is better than well said." - Benjamin Franklin

      Hi flexvault,

      Thanks for the performance tuning tip, I will look into that some time. Though my server is not likely to be used heavily with high traffic. But performance checking it might be good for finding any potential memory or resource leaks or high CPU consumption, deadlocks. Otherwise, I don't really need to optimize the code for performance at this time.

      I actually made this post in terms of code design improvements. Specifically 2 things. (1) whether some of the code could be written according to modern (and proper?) Perl standards and using better Perl modules, optimizing for standards (rather than performance). (2) for any suggestions on how to fix or tackle the open issues and todo items I've listed in the project.

      FYI, the server is used to serve QA/automation test libraries written in Perl remotely (on different machine or even just localhost) over XML-RPC for integration/use by RobotFramework.org test automation framework, using it's remote library API specification (which utilizes XML-RPC). The details on how the server works and what it's use is can be found on the wiki page link I mention earlier and in the Robot Framework documentation. Functionally, it works for basic support, but there's a few areas that can be improved for it to be up to standards with the reference spec implementation (the Python remote server for same framework).

Log In?
Username:
Password:

What's my password?
Create A New User
Node Status?
node history
Node Type: perlquestion [id://1010245]
Approved by davido
help
Chatterbox?
and the web crawler heard nothing...

How do I use this? | Other CB clients
Other Users?
Others chilling in the Monastery: (10)
As of 2014-08-28 11:31 GMT
Sections?
Information?
Find Nodes?
Leftovers?
    Voting Booth?

    The best computer themed movie is:











    Results (259 votes), past polls