Hi Niels, thank you for getting the discussion started! Couple of initial thoughts :
1. I think that ytd-meego uploader/client (2) could interface with the existing youtube-direct implementation. This will make it attractive for folks in the community who have already deployed YouTube Direct and want to use it with MeeGo devices, this page has some examples (scroll to the bottom of the page) : www.youtube.com/direct . Therefore, form my perspective the near-term prio of meetube (1) as a replacement/port of youtube-direct is not clear, unless I misunderstood the purpose of meetube. 2. Wrt the YouTube API itself, we currently have multiple language bindings (see the list here : http://code.google.com/apis/youtube/code.html), but I'm not sure if they're adequate for MeeGo developers. The good news is that the APIs can be accessed using straight REST/HTTP, and this is, in fact, what ytd-android currently does. Thanks, Jarek www.twitter.com/wjarek On Wed, Nov 3, 2010 at 11:26 AM, Niels Mayer <nielsma...@gmail.com> wrote: > A few weeks ago, Quim Gil posted a request to port Youtube-direct to > some meego lists and IRC: > > Qt Quick developer wanted to develop openly a port of > > http://code.google.com/p/youtube-direct/ - documenting the experience > > dealing with YouTube APIs and Android code. The prize will be an N900 to > > test and a dose of fame in the MeeGo, YouTube and Google developer > > communities. > > As I lack a N900, and can apply my experience in the myriad divergent > underlying technoligies employed in the youtube-direct, I'm putting > forward my intention to lead this project, which is clearly more than > a "one person" project in scope. In return, I'm hoping to gain some > practical experience in using QtQuick/QML and developing "hybrid" > web/media handheld apps on Meego. I'm working on a java-based > web-application which will eventually make use of Youtube API's ( > http://nielsmayer.com/trainspodder-prototype.jpg > http://nielsmayer.com/ts-episode-timeline.png ) and for which I intend > to build a handheld interface in Meego employing both AJAX-style > JavaScript Browser-Based UIs, combined with native linux > functionality, to enable tightly-synchronized audio/video&hypermedia > -- both for playback and upload. > > My initial analysis is that porting > http://youtube-direct.googlecode.com for Meego will be more complex > than necessary if we take a direct path in creating something > competitive with http://ytd-android.googlecode.com and > http://ytd-iphone.googlecode.com . There's a certain > Rube-Goldbergification that happens when you split up your apps AJAX > style -- see the considerable amount of setup needed to get the > google-app-engine part working ( > > http://groups.google.com/group/youtube-api-gdata/browse_thread/thread/a67cd4a41692a6ec/f9f40268137bc742 > ). Additionally, there's the dependency on Eclipse IDE, which becomes > impractical given that QtCreator is Meego's IDE. Ultimately, all the > "java cloud" gives you for this app is performance problems that > cannot easily be resolved: > ( http://groups.google.com/group/youtube-api-gdata/msg/ada27fc26a08b78e > ). And the "cloud" is totally unnecessary, since Java is performing > computations that could easily be carried out by a Qt4 app running on > a handheld using traditional REST web programming style. > > The first question to ask in such a project is whether youtube-direct > is even solving the right problem given the strength of the client > platform provided by Linux and Qt? > > Is this "java cloud" approach even necessary, or is it a "wart" > associated with AJAX style UIs? IMHO simple REST ( > http://en.wikipedia.org/wiki/Representational_State_Transfer ) should > handle it. For example > http://wiki.forum.nokia.com/index.php/QtMobilityYouPlayer makes use of > the Youtube API, but doesn't require a app-engine back-end to host the > application. Unlike AJAX-style programming, the decision of how much > processing can be done locally versus "in the cloud" isn't already > pre-chosen for you and the application architecture doesn't end up > being twisted to suit the security limitations of the client. Other, > much-simpler implementations of Youtube-direct exist which don't need > google-app-engine at all, e.g.: > > http://groups.google.com/group/youtube-api-gdata/browse_thread/thread/670c6a0867ca3852/45e97f0feb2e7244 > ("Youtube Direct Upload with clientlogin :: PHP without client library") > > http://groups.google.com/group/youtube-api-gdata/browse_thread/thread/280115751a3ba86/cd76b66db7e797ce > ("Delete a playlist with PHP (You must specify an URI to which to > post) Options"). > > If we do go with the approach of a "direct" port employing all the > browser-based AJAX code, then does it make sense to use > http://wiki.forum.nokia.com/index.php/Hybrid_Application_Generator ? > It's clear this kind of AJAX-based application needs an integrated > javascript-enabled browser (QtWeb) and additional code support for > "platform natitve" audio/video uploading, audio/video playback, > GPS/geotagging etc. Alternately, I've also been told to investigate > the QML bindings for QtMobility Media APIs. > > For "baseline functionality" lets consider what "Android YTD mobile > uploader" does: > ( > http://groups.google.com/group/youtube-api-gdata/browse_thread/thread/17020651ef1aa52b/64a0a0a9031766d2 > ) > * Integration with Android's Account Manager for ease of authentication > (the > application prompts the user to select a configured Google / Gmail account > on the mobile device) > * Automatic synchronization of assignment list (submission ideas) from the > YouTube Direct JSON assignment feed > * Notification display upon discovery of new assignments > * Video recording and upload to a specific submission idea > * Upload of a video selected from the gallery > * Geolocation tagging of submitted videos > > ------- ------ ----- ----- > > As a starting point around which to organize these issues, discuss > features, and implementation strategies, I propose to start a project > on > http://googlecode.com for porting the features and functionality of > YouTube-direct to Meego. As I believe the scope of this effort is > larger than any single person can provide, it makes sense to start it > as an open-source project and seek contributions from the community. > > First order of business is the name: > > (1) http://meetube.googlecode.com > > -- My preferred project name: memorable & short & humorous. > -- Since project ultimately may not take "direct" approach used by > youtube-direct, don't want to name it "youtube direct" > -- maybe someone from marketing won't like it or it violates some > trademark on "mee" or "tube" :-) > > (2) http://ytd-meego.googlecode.com > > -- parallel name to http://ytd-android.googlecode.com and > http://ytd-iphone.googlecode.com > -- but may not even want to use "youtube-direct" approach and just > make use of Youtube API from Qt4. > > With the name chosen, the next step will be to generate a set of > issue-tracker feature requests ("Type-Enhancement") -- one per major > feature > of "Android YTD mobile uploader" app mentioned above. Also, will > create issue-tracker requests ("Type-Review") to organize issues, > positions and arguments for/against implementation strategies: (i) > "hybrid application" employing QtWeb; (ii) QtQuick/QML ; (iii) > Standard Qt4 app that issues and responds to web-requests without > needing a "browser", e.g. > http://wiki.forum.nokia.com/index.php/QtMobilityYouPlayer ; (iv) A > more direct port of "ytd-android" or "ytd-iphone." > > Please send mail if you want to be added as contributor or committer > on the googlecode project. > > Niels > http://nielsmayer.com >
_______________________________________________ MeeGo-community mailing list MeeGo-community@meego.com http://lists.meego.com/listinfo/meego-community
