Showing posts with label code. Show all posts
Showing posts with label code. Show all posts

Wednesday, April 11, 2012

Bloated Desktop Applications


While reviewing my options for caching some graphic containers in Curl, I checked my process explorer.

My browser was creeping up to 1.8 GB as a busy, greedy process on an XP Pentium with 1 GB RAM.

The browser is not the only offender: otherwise useful Evernote appears to lack some smarts (remember when Prolog simply required too much memory?)

Let the OS worry about it?  Throw more memory in the device?  Ascend to the free cloud?

Is there now a generation of Perl/PHP scriptors working in HTML + JavaScript for whom memory and CPU cycles swapping-to-disk are free? Are they avoiding multiple trips this way?  Is this the AJAX heritage?  SSD merely warm, but not whirring?

Just as it appears that Rebol may disappear.

Remember why Smalltalk was not in use (outside Amex and JP Morgan and ?)  St required a VM and too much memory.  Now as some Smalltalk implementations shrink and their VM's enter a new generation, ask a corporate programmer when she last prototyped an app in a Smalltalk dialect.

I can hear it ringing in my ears: "All we need is one little Perl script to do that !"  But CGI with Perl was too costly on servers, remember?  But now it appears that nothing is too costly on a client device.

But free memory, like free DASD, is only as believable as free money, or 33% annual return on your investment if you get in early.

Where is the desktop UNICODE-savvy text editor app that knows how to edit multiple small files without burping up a 30MB or more process for each tiny text file?

These apps are the bungalows of the sprawling suburbs of 50's Middle-America - and now they propose to bloat invisibly within the cloud. Is this a viable global picture of net resource consumption?  Which economic model suggests greater bandwidth as evolution's competitive winning variant?

Maybe kids running Rebol on Raspberry Pi knock-off's will get us on another track.

Try this: compare some smart e-commerce sites to the UI's of Google PLAY or Canada'a zip.ca

Those UI's are not JavaScript generated by Smalltalk - or by Prolog.  These sad web pages must each re-evolve from dumb and primitive.  These are not smart lean app's standing on the shoulders of bulky slow app's. That is just not how the web is evolving.

PS
Ruby is just a dumbed-down Smalltalk for Perl scripting.

Scala ?  Let me tell you a story that is not up-Lift'ing ...

Multi-core?  Watch for MC Smalltalk.  But bloat and greedy prcesses remain what they are just as a Ponzi-scheme by any name remains your personal goldmine .... and your cousin's, too !



Friday, July 3, 2009

Aule Browser arrives at code.google

There is now an opensource Aule Browser project on code.google.com

Areas to watch are generating Aule browser wrappers with the Seaside web framework and other templating and domain-specific language (DSL) and dialect issues for generating site-specific browsers (SSB) in the Curl programming language.

The Aule browser uses the Curl EmbeddedBrowserGraphic class and its related EmbeddedBrowserEvent class in a detached Curl desktop rich internet client application.

Curl is a mature and established client-side enterprise platform but as a web content language, Curl remains an alternative to HTML + JavaScript with separation from presentation style.