[Seaside] My progress with Seaside...

Wed Jan 4 18:42:50 CET 2006

Jeremy Shute wrote:
> My progress with Seaside has been impeded by a few things.  It might be
> helpful to others to try and "troubleshoot" these on the mailing list.
> 
> * <frameset>.  Right now there's really no good support for showing things
> in frames.  I had to override WARenderLoopMain, WARenderLoop,
> WARenderedHtmlRoot, and WARender in order to simply replace the <body> tag
> with <frameset>, and once I have done that, there is still no support for
> targetting different frames with callbacks and events.

Yes, Avi and I never liked framesets (I always thought they were clunky 
and ugly) so we never used them.  I think there's IFRAME support those 
(or there was at one point) and with that and ajax, frames seem even 
less useful than they once might have been.  <shrug>  Not that we want 
to prevent others from using them, just that nobody's cared enough to 
put in support for it and as you have seen it permeates a lot of other 
code to do so.

> * Speed.  The profiler is a great tool -- it's telling me my page takes ~3
> seconds to render.  About 80% of the time is spent on a component that is
> being repeatedly rendered.  You can see the TagBrush code chugging:

<snip>

 > The API is beautiful (and the cascading solves a lot of the issues I had
 > using the old API).  However, I cannot get my pages to load fast enough.
 > My goal is for any page to load in under a second.  There doesn't seem to
 > be one glaringly obvious region for speed improvements...  What can I do?

Sorry... no time so stare at the profiler this morning.  How big are 
your pages?  The surest way to reduce render time is to reduce page size 
(or get a faster machine :) ).  I usually try to keep content to a page 
or two.

> * Event propogation.  I can't figure out how to have an "href" that goes
> somewhere in a text browser and an onClick event that returns "false" to
> stop it triggering if Javascript is enabled.  Manually setting the "href"
> attribute stops the "javascript:void(0)" behavior, but is there an easy
> way to get the event to stop its propogation, such that both Javascript
> and non-Javascript behavior can coexist?

Not sure on this one.

> * Bandwidth.  Loading up the SUScriptLibrary for AJAX support, I get
> another 120K worth of stuff that I just plain don't need.  Yes, 95%+
> browsers support AJAX, but half the country is still on a modem!  The
> whole point of AJAX is to make a snappy application!  I see two options to
> fix this: meet my own requirements by returning to the V2.5 way of doing
> things (i.e. manually port the much slimmer 2.5 AJAX code), or write
> Javascript dependency graph analysis code that only sends what you use
> (which is a nightmare without some additional declarations).  REALLY, I'd
> like an "AJAX lite" library that has all the handlers but none of the
> animation: is there a way to get 2.5 functionality back?

Is half the country really still on a modem?  I made some comment about 
supporting modems for online education the other week and my employer 
scoffed and said common wisdom was that nearly everyone had access to 
high-speed now.  I just pulled up a couple of #1 hits from google on 
statistical queries and (very roughly):

http://www.oecd.org/document/16/0,2340,en_2649_34225_35526608_1_1_1_1,00.html

says that in the US there are 14.5 broadband subscribers for every 100 
inhabitants.  Taking the average family size of 4.4 (is that still what 
they use?) that's 14.5 subscribers for every 22.7 families.  So 63.8% of 
people are in families that have broadband.

According to http://www.internetworldstats.com/stats2.htm

Internet penetration in the US is 68.7% of the population.  So that's 
more like 93% of US internet users have broadband.  Now... that doesn't 
account for there being very young children counted in the families who 
probably aren't counted in the internet penetration and the broadband 
subscribers may include business subscriptions (although those users 
also arguable have access to broadband at work so that might skew the 
data in either direction).

I'm not claiming it's terribly scientific but I'm curious if anyone else 
has any real data on this.  Interestingly, broadband penetration is 19.2 
per 100 in Canada which is 84.5% of people who have access to broadband 
if our family size is the same.  Ok, and internet penetration is 63.8%. 
  So is it possible that nearly 20% of people who have access to the 
inernet don't use it?  Between very young children and some family 
members probably not using it (and again allowing error for business 
subscribers skewing the numbers in both directions) I'd say that seems 
possible.

Anyway, amateur statistician hat off... off to work (agh, I'm late!)

Julian