Re: [Re: [Dynapi-Dev] compression?]

[Bill W. <bil...@us...>]

I know it would take a better programmer than I am to do this.
I had a program that would syntax parse a js file, and replace all variable
names with aliases. For example, 'DynDocument' might become 'aXn' etc.  Next,
it takes things like this:
 if (fooBar2){
  doOneLine()
  alert('hello')
 }else{
  doAnotherLine()
 }
 var x = 10
to:
if(xTa){bFg();alert('hello')}else bFh();var x=10

Note, that it puts in the semi's by itself, renames functions to shorter
names, gets rid of senseless braces for one liners, gets rid of all comments,
and allows you to add copyright comments as an include when you run it.
http://www.geocities.com/SiliconValley/Vista/5233/jmyth.htm

-bw

Dan Steinman <dy...@fu...> wrote:
> Pascal, initially I wasn't so hot to trot with whitespace compression
either, but after playing around with it it's really quite cool, and I'm glad
I put in the time to create my little jspack util.  If you check again, the
compression ratio is closer to about 30%, more or less depending on how many
comments you put in there.  All the /api/ files were "manually" compress.  I
don't particularly enjoy reading code that is like this:
> 
> if(b)c=2
> else if(a&&b||c&&d<=4)method()
> else {method();return false}
> 
> And with the system set up now, we won't have to.  You just write your code,
putting in all the comments and whitespace you want, and (as long as the ;'s
are correct) it will all be squished down for you in a build release.
> 
> Also, jspack compresses stuff that gzip/jar's can't do much with, like the
following:
> 
> if (a) {
> 
> }
> else if (b || c) {
> 
> }
> 
> The spaces and line breaks have to be maintained.  But in a jspacked
version:
> 
> if(a){}else if(b||c){}
> 
> It's already taken care of.  After you jspack, and jar/gzip it, you're up to
over 80% compression (85% for dynapi2 currently).  Whereas gzip on it's own
can only do about 60%.  Plus by whitespace compression, each individual file
can be compressed, whereas there's no point in having an individual gzip/jar
file for each js file because you don't gain much in small files. 
> 
> Another thing that's kinda cool, imagine when you build a game, or
standalone app in JS.  When you want to release the app to the public, you can
build your own JS pack file with the code you need all in there (GPL license
allows u to do this if the program you wrote is also GPL).  I'm rewriting my
tetris game, so I'd make a jspack for it with:
> 
> jspack -o tetris-packed.js tetris.js ../dynapi/dynapi.js ../dynapi/api/
../dynapi/gui/image.js
> 
> That produces a single js file with all the code I need squished together. 
This is an ideal way to publish js apps or games IMO because you no longer
need to include the entire dynapi with your app (or copy select files along .
> 
> Dan
> 
> 
> On Fri, Dec 08, 2000 at 12:49:06PM +0100, Pascal Bestebroer wrote:
> > Not to be raving on anybodys parade, but whitespace compressions isn't
> > really worth it is it?  You only stripped about 4kb off the complete api/
> > folder (dynlayer,dyndocument,etc..) if you remove all comments and
> > semi-colons from the un-"compressed" files, you will probably end up with
> > even less "compression".
> > 
> > I know that compression on all widgets might give more advantage, but
then
> > again when will you be actually using all widgets?
> > 
> > I think the jar and gzip are great, but jar won't work in IE, and I don't
> > think gzip works in ie4 (and how about IE on Macs?)
> > 
> > Am I missing some advantage in this?
> > 
> > 
> > 
> > Pascal Bestebroer
> > pb...@oi...
> > http://www.oibv.com
> > 
> > _______________________________________________
> > Dynapi-Dev mailing list
> > Dyn...@li...
> > http://lists.sourceforge.net/mailman/listinfo/dynapi-dev
> _______________________________________________
> Dynapi-Dev mailing list
> Dyn...@li...
> http://lists.sourceforge.net/mailman/listinfo/dynapi-dev


____________________________________________________________________
Get free email and a permanent address at http://www.netaddress.com/?N=1