[Urwid] A couple of urwid-related talks I'm planning to give

Rob Lanphier robla at robla.net
Sun Apr 4 17:22:48 EDT 2010

On Sat, Apr 3, 2010 at 7:47 PM, Ian Ward <ian at excess.org> wrote:
> The twisted-conch work Ali Afshar started (serve lots of ssh users with
> one app) is very promising, very cool.  I hope to pull that in to Urwid
> some time soon.
> Also, the work to support Twisted and Glib by just changing a parameter
> on the MainLoop class is extremely handy for apps that need it.

I'll make sure I at least briefly touch on those.  Would it be fair to
say that the main benefit of a GLib event loop is that it makes it
much easier to pull in the glib bindings and do things like listen for
d-bus events?  What's a good example of a theoretical but practical
application you would use to get people excited about using a GLib
main loop?

> And maybe something about Urwid not trying to be TurboVision (for those
> old enough to remember that)  UI concepts that work well on the console
> are often different than ones that work for GUIs.  The most popular
> console apps are space-efficient and keyboard-focused.  I've tried to
> write Urwid for that kind of app.

That's good to know.  I'm going to have to think about delivering that
message in a way that doesn't conflict with the message I've been
thinking about.  I'll definitely make it clear that you aren't trying
to make this into TurboVision, but part of what interested me in
giving this talk is to encourage people to write applications that are
more immediately intuitive than many of the console apps that are
popular today.  I don't think what I'm saying is at odds with your
point, but let me see if what I'm thinking about here rubs you the
wrong way.

I think up until recently, the payoff for investing in console apps
has been pretty low (i.e. 8 hours of ncurses programming doesn't get
you much of a user interface, and there's little that you get "for
free").  Now that it's not that hard to knock out a reasonably looking
user interface, there's no reason why the distros shouldn't make
console usability part of their general usability push, especially
since good console apps would probably be most useful to the big
distros highest paying customers (Linux servers account for *way* more
revenue than Linux desktops, and "the cloud" is shaping up as the next
big battleground).  Usable console apps are much easier to create now,
so why not actually try to make more of them, and make the ones we
have more usable?

As I think you're saying here, though, "usable" doesn't necessarily
mean ASCII-art borders, drop shadows, and a mouse-centric UI.  People
using SSH may very easily still be constrained by what's possible over
vt102, so making a lot of assumptions about I/O devices and screen
real estate may easily result in something less usable.

Is that compatible with what you see as the vision for the urwid project?

> Finally: pretty colours.  Urwid supports lots of them.  :-)

Ah yes, the colors!  I'll make sure that's obvious.

> BTW, Are there going to be videos of either of the talks available?

I'm not sure.  The last time I was at LFNW (in 2008), there were some
rooms that had video equipment, and some that didn't, so it depended
on where you were.  OSCON seems to typically only put videos up of the
keynotes, so I don't think that one is going to be available, but now
that video is getting so cheap and easy to shoot, maybe this year
they'll expand out.


More information about the Urwid mailing list