As far as OOP, I agree that something like object inheritance is for humans, not the computer. Which would make a script based "class" hierarchy completely burdensome to the server. It would have to do all the lookups as you said. But if it is compiled, the compiler does the lookups at compile time, and optimizes the app, and the finished app does not suffer from having to do all the lookups. Isn't that the whole point of compiling? Converting a usable human language into optimized machine code?
Robert.
On Monday, June 30, 2003, at 08:09 PM, Atrix Wolfe wrote:
if you want compiled speed you can always "compile" your code to java
bytecodes. If you want TRUELY compiled speed, gcc (which is ported to many
many platforms) has the ability to take java bytecodes and make them into
exes or dlls. You cant get much faster than that.
IMO (im sure others disagree) i think the whole idea of OOP leads
programmers to make programs which make sense to them, but dont make sense
inside a computer.
For instance, you might make some classes like an apple is inherited from a
fruit, which is inherited from a plant, which is inherited from an entity
which is inherited from an object....
makes sense to us, however to reference an apple, it would have to go
through a whole bunch of lookups just to get to the data it needs. If you
could see your computer working inside somehow it would be jumping all
around in RAM just to get to one spot that it could have gotten to much
quicker if it went straight there, not to mention all the memory overhead to
store all the lookup tables.
while it is useful to an extent in abstracting ideas and data, i think OOP
is often misused and often times gives the wrong impression to people making
programmers about what is or is not efficient.
with modern computers this isnt really a huge problem as fast as they are
but if your worried about preformance, I thought this might be relevant (:
my 2 cents! Atrix
----- Original Message ----- From: "Robert Garcia" <[EMAIL PROTECTED]> To: <[EMAIL PROTECTED]> Sent: Monday, June 30, 2003 8:01 PM Subject: Witango-Talk: Boosting Server Performance, and TCFs
One of the greatest flexibilities in Witango is that it is a scripting language vs. a compiled language. It allows for rapid application prototyping and development, and the ability to do incredibly flexible operations like dynamic object referencing.
However, a compiled language has the benefit of performance. It doesn't
have to read/parse all the tags on the fly, and then perform its
function. It just accepts input and performs its function.
Language like perl, python,php, are purely scripting langues.
Others offer both under one roof. Like the JSP/J2EE model. JSP pages are script based, but make calls to compiled java servlets for performance.
Witango does offer external actions, and java bean support, and will be
offering the ability to output to JSP, but I would love to see a
tighter integration with compiled performance. Also, JSP allows the use
of another server, and remove the need for the Witango server, which
probably isn't the desired use of this tool.
I would love to see Witango develop a tighter integration with compiled
performance. Currently, there are three file types to use when creating
witango apps, the tml, taf, and tcf. In my opinion, the tcf is the
first place I would see a great benefit in compiling before serving.
As far as I understand now, using TCF's actually reduces server performance. It is great for separating your code, and creating reusable code to use on multiple apps, but there is at best, no performance advantage, and I believe it is a performance hit. I have seen many witango developers completely abandon tcfs due to the extra learning curve, and when they realize it is a performance hit, they just use branching.
But what if the TCF could be compiled for deployment, providing a large
performance increase for code run in a tcf. And I don't mean JAVAC
compiled, but compiled with the same native compiler process used for
the server? It would completely outperform JAVA and give Witango
developers the ability to seriously boost performance in critical
applications.
Granted, this would mean Witango would have to write a compiler, and certain operations would not be allowed in a tcf that are allowed now, but I think this would be a great step.
I am mainly thinking out loud, but I wanted to start a discussion on this topic for a while. I love TCFs, and I love writing code in a more object oriented method, but I hate that it is a performance hit.
Maybe after this discussion, we can talk about making TCF even more
object oriented. Like maybe inheritance and method overloading, and . I
would love to be able to extend a tcf "class" in a child class knowing
that when I update the parent, the children inherit the benefit.
--
Robert Garcia President - BigHead Technology CTO - eventpix.com 2781 N Carlmont Pl Simi Valley, Ca 93065 ph: 805.522.8577 - cell: 805.501.1390 [EMAIL PROTECTED] - [EMAIL PROTECTED] http://bighead.net/ - http://eventpix.com/ - http://theradmac.com/
______________________________________________________________________ __
TO UNSUBSCRIBE: Go to http://www.witango.com/maillist.taf
_______________________________________________________________________ _
TO UNSUBSCRIBE: Go to http://www.witango.com/maillist.taf
--
Robert Garcia President - BigHead Technology CTO - eventpix.com 2781 N Carlmont Pl Simi Valley, Ca 93065 ph: 805.522.8577 - cell: 805.501.1390 [EMAIL PROTECTED] - [EMAIL PROTECTED] http://bighead.net/ - http://eventpix.com/ - http://theradmac.com/
________________________________________________________________________ TO UNSUBSCRIBE: Go to http://www.witango.com/maillist.taf
