Next steps for Witty

[Jon Galloway]

I wanted to give a quick update on Witty to let everyone know where
we're headed and to get everyone's feedback and involvement.

Big picture - Up until this point, Witty has continued to grow
organically through the generous contributions of a lot of developers
who love Twitter, wanted a "pro" client that worked how they thought a
Twitter client should work, and (often) were interested in building
their skills in C# and WPF. Speaking for myself, at least, I've
learned a lot about WPF development and design patterns during the
time I've been contributing to Witty. This is both awesome and un-
awesome.

Awesome: We've built a really cool Twitter client that lots of people
use every day.
Awesome: We've all learned a lot by working through some tricky coding
challenges together.
Not Awesome: Patch Driven Development is a great way to build a lava-
flow architecture.
Not Awesome: The architecture we've got now is holding us back. Since
the Witty project started, some solid design patterns like MVVM have
emerged. These patterns aren't patterns for patterns' sake, they
enable some desirable development features like Blendability, more
flexibility in presentation, possibilities for better information
management, and simplified development.
Not Awesome: Our visual design is functional, but not what it could
be. I don't think Witty's users value form over function, but Witty
could look better.
Not Awesome: Applications like TweetDeck and Seesmic have shown that
there's some value in a multicolumn view. We wouldn't get rid of the
compact view that Witty's got now, but we could better handle the
screen real-estate when we're given more.

Recent developments:
One of the struggles we've had to date has been a lack of interest
from the established WPF community. That's changed lately - we've had
some good input and offers to help. Alan Le and I had a discussion
with Tim Heuer and Chris Bennage at MIX09, and we recorded the
discussion so you can hear what what we were talking about:
http://herdingcode.com/?p=175

Silverlight 3 Out Of Browser was announced at MIX09. There are some
interesting possibilities for a Silverlight-based Witty (e.g. simple
install, cross-platform reach), as well as some tricky challenges
(right now the Twitter API doesn't allow cross-domain access from
Silverlight).

Some next steps:
We've been brainstorming on a Witty 3 architecture. Chris Bennage has
offered to prototype some of that for us based on his WPF ChumChase
app (http://www.codeplex.com/chumchase). Alan and I've both done a
good amount of work with MVVM lately, so if Chris punks out on us ;-)
we'll jump in.

We've also had some good input on UI and offers of help from some
really good UX designers.

Next steps:
Keep contributing to Witty. We should be able to reuse heavily code
and features from Witty 2.x in Witty 3.

Contribute your ideas, either in the wiki (http://code.google.com/p/
wittytwitter/wiki/WittyTwitterRoadmap) or by submitting issues.

Thoughts about changing project hosting:
We've talked about moving to CodePlex, mostly since that would
probably grease the wheels for getting more involvement from the
Microsoft community. Based on an involved discussion with the Subtext
team, I'm thinking it's best to stay with Google Code right now. You
can read Phil Haack's thoughts on moving to Google Code instead of
CodePlex here: http://haacked.com/archive/2009/03/29/subtext-moves-google-code.aspx
What do you think?

[Keith Elder]

I know that Codeplex is going to natively support ClickOnce
deployments, and that is a big win for the project that google code
doesn't or will never support. Today we don't have a good story on
people getting and installing Witty.

I'm not sure when it is coming, but it is coming to Codeplex (if it is
not already there). Personally I've never noticed some of the things
Phil mentioned (speed for example).

I really like the issue tracker in Codeplex and ability to vote and
their are plans to allow you to update pages in Codeplex using Live
Writer (makes it much simpler).

My $.02.

[Dimebrain]

I chose Google Code for TweetSharp for more or less the same reasons
as Phil did. I used CodePlex for all of my projects in the past, and
it was workable, but too slow for me. Though to be fair I used it back
when it had no decent SVN bridge support, if that is fully there at
this point then I couldn't claim it was better or worse than Google
Code.

On a side note, TweetSharp is continuing to keep up with the Twitter
API changes, like some of the nastier recent reversals (deprecating
the since_date parameter for instance, changes how you need to think
about paging tweets, especially since count= will actually return from
the top of a descending ordered list, not what you would expect) ---
disclosure being that we use it to dog food our own WPF / SL3 Twitter
client application intended for sale / non-open source use --- we're
more than happy to consider patches, and you will at least know that
someone is staying on top of the API behind the scenes.

Also, I am willing to help update the source when things as drastic as
since_date change, or hook up TweetSharp behind your controller /
threaded task classes. Maybe the equivalent of the grease monkey who
works on the starship. :)

I'm looking forward to seeing the next generation of Witty.

[Stefan Dobrev]

Guys, have you considered using prism for the next version? It implies VM pattern and also will be a good step forward to the Silverlight version.

2009/4/18 Jon Galloway <jonga...@gmail.com>

[dweinand]

what's about github (and lighthouse)? i think for distributed
developing and getting people aboard there's no better choice at the
moment.
just let them fork away and take all the "patches" back as simple as
123. it's much easier to get started.