On Saturday, May 4, 2002, at 02:26 , Greg Willings wrote:
> Hi, > I'm a beginner that doesn't even have perl yet. > I woul like to know whether Perl is faster or Java for business > applications. Paul has already provided the correct Party Line from the typical Perl Advocate. So I'll be stuck trying to defend the grey beards position. As for KH's variation on the theme, now we start to move towards the place where we have some parameters to really work with. To save some space I will try to 'speak in uri' - One of the places I have been ranting on this is: http://www.wetware.com/drieux/CS/SoftEng/GP/funcOrNotFunc.html Where I discuss the matters of "code reuse" - the OO process by means of 'class/sub_class' - in Proceduralist languages this is the library - in Perl it is the Module - the problem in perl is that the perl5 OO implementation is not as good as it can be, perl6 may 'solve all of that' - and we will see at what costs. If you want to be way trendy then you of course want to be looking at the Objective Caml Language: cf: http://www.ocaml.org/ Since this is a language where first order functionality exists and it allows for the use of OO in the few possible edge cases where that may actually have some 'requiredNeff'. Assuming that one were truly seeking 'ubiquity' for 'human interface' then of course one has already come to understand that one is either going to be writing CGI or mod_perl and/or apache modules - since of course no 'real Operating System' can exist but that it is fully integrated into a webBrowser. { and we thought the emacs folks were a bit over the edge. } At which point one is really talking in terms of what specific type of Tomcat Servlet v. mod_perl v. apache module were you really planning to write.... { old guys of course will be thinking in terms of Perl/Tk - because they do not understand that Operating Systems that are not a part of a webBrowser are passe.... and still have this irrational belief that there were kernels that were not browser plug-ins. } { for those of you new to writing device drivers - please, do not assume that simply because your VB based device driver allowed you to use 'standard IO' reads of flat files for configurations - that this will allowed in an adult Operating System - and no, you can not put 'printf()' or 'fprintf()' statements in to do your kernel debugging. Nope - Not an option. } As for the specific question of Large Database - we of course must pause and ask whether or not the system is a 'data mining' project where it will be used for secondary 'decision matrix analysis' - and speed is not the issue - or are we talking in terms of an OLTP - at which point the real question is who is doing the database design and shielding the running daemons from the DB itself by doing the appropriate "stored procedures" and transaction abstractions - at which point we're also wondering why you want to do the API to that in anything but 'c' because you really want to have the most portable 'assembler code' processing system that works from statically linked compile time code re-use - because you are not at all interested in engaging in the run-time dynamically loadable link level library hassles of version skew as the libfoo.2.so really is not backwardly compatible with the libfoo.1.so and you are not too sure how to fully back out all of the code that needs to use the old form since the process installed them as libfoo.so -> libfoo.<int>.so and there can only be ONE! At which point we could get into the discussions about whether polymorphism is a good idea, as well as whether or not java - which expressly avoids multiple class inheritance - fully got away from all of that with the 'implements' that allows the maze of twisty little passages as you find that there are 'structural' issues getting the extends stuff to work and play well with the implements stuff.... We should of course write up the problems associated with the whole process of 'kargo kulting' - In which the blind pass along what they heard to help the deaf find a way to smell a fart... Which has more to do with whether or not you personally should be wearing the trendy T-shirt that says: "WARNING: does not work and play well with others." Especially if you keep hacking out work arounds for what the 'realCoders[tm]' were supposedly 'developing' and finding that this is so much more pathetically stoopidly implemented and maintained in perl than in one of those 'real coding languages' that are the topic of hotBuzz. Hence - just as you should have learned at least one assembler language to know why you write that stuff in 'c' - you need to know Perl so that you know why you don't write 'fast code' in an object oriented programming paradigm - since you really do not want to wait around while it bloats up reimplementing in each object a common set of methods for all gets and sets so that it can be fully encapsulated... Or because you find that hey, speed is not the need, hence why not just use this Module - rather than re-implement it as a flat stack of functions.... So that you can get on with the truly important part of your life - impressing your friends and neighbors in the various forms of DickSizeWars[tm] that are your TechNoir moment as you still haven't quite figured out how to hack access to a gender appropriate person - because they don't come with any of the standard networking protocols.... But aren't they suppose to have at least an RS-232 port? ciao drieux --- -- To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
