> -----Original Message-----
> From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED] On Behalf Of 
> David Nuescheler
> Sent: Monday, June 09, 2008 3:19 AM
> To: [email protected]
> Subject: Re: JCR browser
> 
> hi craig,
> 
> i think this is a great start. congratulations.
> there are a couple of observations that i made which are 
> likely to be unrelated to the use dojo so i thought i'd share 
> them here.
> 
Thanks, I appreciate the great feedback!

> (1) deep links, bookmarks & backbutton
> since i am fan of deep links into content when i am sharing 
> content locations i think it would be a great feature to 
> expose the current navigation path in the url that's 
> displayed in the browser.
> this can for example be done through a #<path> in the url 
> basically whenever i click on a folder in the tree.
> this should also fix back button and bookmarkability.
> 
Ok, let me think about this, it sounds like a good idea.  Note that I
have spent *0* time on the URI scheme for this yet ;-)

> (2) cookies
> it seems that all the opening states of the tree are stored 
> in cookies... i would probably only store the last open 
> position in the cookie since in my experience it is sometimes 
> even more desirable to have sort of a "clean" start, with 
> only my last position reopened ;)
> 
Yeah, this is something that dojo is doing for me automatically and I
hadn't thought about how best to use the feature yet.

> (3) safari works beautifully...
> ...but complains about a slow script 3 times before the 
> browser starts up.
> i think this could be addressed by looking into what takes 
> the load time.
> in firefox with firebug enabled for tracking the startup of 
> roughly 43s which is probably most network latency related.
> 
When I'm ready for a final release, I am planning on doing a custom
build of dojo which *should* address this issue. I believe all of the
startup time is attributed to getting all the .js files that are
required, and I plan on building dojo so that it's just one .js script
with only necessary requirements.  That should help initial startup
immensely.

> (4) network roundtrips
> i think that once the browser is up and running the amount of 
> network traffic that is transported is in very good shape.
> i stumbled over an incident that surprised me a little bit though.
> i expected a single .json to be transported to the client 
> when i would open a new tree node clicking on the [+], which 
> is true for most nodes, but not for nodes with deep children 
> underneith... for example opening /images issued 9 http requests.
> 
Yes, there are a couple of things that probably contribute to this,
first dojo's idea of a lazy loading tree doesn't fit with how *I* would
do it.  I'd developed a store that did real lazy loading, only load the
children when a node has expanded, but I haven't tested it against 1.1.1
(it did work with 0.9, so in theory it *should* work here too).  I'm
going to be implementing that code in Lars' stores for the JCR
explorer's use.

> i hope some of this may be helpful...
> 
Yes, very!  Thanks!

> regards,
> david
> 
Cheers,
Craig

Reply via email to