|
|
Re: Rap on mobile devices [message #70574 is a reply to message #70250] |
Thu, 17 January 2008 23:03 |
Eclipse User |
|
|
|
Originally posted by: jkrause.innoopract.com
Hi Jorge,
The typical size of the first response to the browser is about 175K. It
is compressed if the browser supports zipped responses, which almost all
browsers do. If the browser does not support zipped responses the
initial size is about 1 megabyte. Subsequent responses are usually very
small (50-2000 Byte).
Looking at RAP and mobile devices there are a few things to explain:
- qooxdoo (with the features we need) has a memory footprint that is
not ideal for devices.
- running a complete workbench on a device leads to a high complexity
on the client side, which consumes resources but provides only little
value due to the restricted screen size.
But if you restrict yourself to some extent RAP should work for you:
As a first step you may want to try running a simple RWT-based app on
your mobile device. I have seen RWT-applications running on Nokia phones.
You can also remove the functionality from qooxdoo that you don't need
(e.g. Tree and Table), this will lead to a smaller client side
footprint. You need to make sure that the corresponding RWT classes are
not being used anywhere. See
http://qooxdoo.org/documentation/0.7/custom_builds and please use the
sources at org.eclipse.rap.rwt/js as a starting point.
If RWT works on the targeted mobile devices you may want to have a look
at eRCP from a conceptual point of view. It provides all the benefits of
modularization, but not the complexity in the UI. I think it shouldn't
be a very big deal to create an "eRAP" based on the eRCP implementation.
Mid-term we will switch to qooxdoo 0.8, which will offer a smaller
footprint, and probably also the possibility to create very lightweight
and targeted UI components close to the browser DOM. DISCLAIMER: I am
not saying that the RAP team will work on those topics soon.
We also plan to move the rendering to a separate layer (fragment), so
that it is possible to use different client side frameworks / technologies.
Please don't hesitate to ask more questions on this forum, RAP on mobile
devices is absolutely a topic of interest for us.
Jochen
----------------------
Jochen Krause
RAP project lead
Innoopract
http://innoopract.com
Jorge wrote:
>
> Hi,
>
> I'm here again,
>
> I have found a way to compress all the libraries javascript before
> loading the application.
>
> manager.register( libraryName,
> HTML.CHARSET_NAME_ISO_8859_1,
> RegisterOptions.VERSION_AND_COMPRESS );
>
>
> The idea is to minimize server-client traffic on application load but
> this mechanism of compression is not used.
> I've tried setting RegisterOptions to VERSION_AND_COMPRESS in debug mode
> and I found that the result was no very good, application size was
> almost the same.
>
> Will VERSION_AND_COMPRESS option be available soon?
>
> There's any other way to get a smaller code size in client side?
> I'm trying RAP in a mobile application but not all generated code is
> necessary.
> There are libraries, comments and blanks not needed. I want to minimize
> loading data traffic.
>
> Jorge,
>
|
|
|
|
|
Re: Rap on mobile devices [message #72593 is a reply to message #72170] |
Fri, 01 February 2008 12:20 |
Eclipse User |
|
|
|
Originally posted by: fappel.innoopract.com
Hi,
a patch for this will be very useful - but are you sure about the bug
number? It leads me to a JDT Core bug - forgive me if I'm writing nonsence
here, but...
Ciao
Frank
"Paul Norrie" <pauln@ablazesoftware.co.nz> schrieb im Newsbeitrag
news:576aea8a27115adbb71d7cbd297d4302$1@www.eclipse.org...
> Hi Jorge
>
> We've just got RAP going on an HP iPAQ running WM05 and a Metrologic
> Optimus PDA running CE.NET 5.0. The browsers used were Pocket IE (WM05)
> and IE6 for CE (CE.NET 5.0) plus Opera 8.65 for WM05.
>
> We also initially got the blue screen only. It turns out this is due to
> javascript errors. Do turn on the ability to view script errors in your
> browser. If the browser doesn't have that option, check out
> http://channel9.msdn.com/wiki/default.aspx/MobileDeveloper.D isplayScriptErrors
> for more info.
>
> We discovered OutOfMemory errors on one device (it only had 64MB memory
> with about 18MB free to use).
>
> On the devices I mentioned above we discovered that qooxdoo tries to call
> window.external which is not supported in any of the above browsers on
> CE/Windows Mobile.
>
> I'm just about to post a patch for this to the RAP team - see Bug 217078.
>
> If I find out any more tips, I'll post them on this newsgroup. It'd be
> great if you could do likewise.
>
> Cheers
> Paul
>
|
|
|
Re: Rap on mobile devices [message #72652 is a reply to message #72170] |
Fri, 01 February 2008 13:00 |
Eclipse User |
|
|
|
Originally posted by: rherrmann.innoopract.com
Hi Paul,
thanks for exploring RAP on mobile devices. I opened a bug at qooxdoo (
https://bugs.eclipse.org/bugs/show_bug.cgi?id=217018 ).
I think there was a typo in you previous posting. The bug id is 217018.
Cheers,
Rüdiger
Paul Norrie wrote:
> Hi Jorge
>
> We've just got RAP going on an HP iPAQ running WM05 and a Metrologic
> Optimus PDA running CE.NET 5.0. The browsers used were Pocket IE (WM05)
> and IE6 for CE (CE.NET 5.0) plus Opera 8.65 for WM05.
>
> We also initially got the blue screen only. It turns out this is due to
> javascript errors. Do turn on the ability to view script errors in your
> browser. If the browser doesn't have that option, check out
> http://channel9.msdn.com/wiki/default.aspx/MobileDeveloper.D isplayScriptErrors
> for more info.
>
> We discovered OutOfMemory errors on one device (it only had 64MB memory
> with about 18MB free to use).
>
> On the devices I mentioned above we discovered that qooxdoo tries to
> call window.external which is not supported in any of the above browsers
> on CE/Windows Mobile.
>
> I'm just about to post a patch for this to the RAP team - see Bug 217078.
>
> If I find out any more tips, I'll post them on this newsgroup. It'd be
> great if you could do likewise.
>
> Cheers
> Paul
>
|
|
|
|
|
|
Powered by
FUDForum. Page generated in 0.03405 seconds