howdy, Suzanne!

hi mark

Am I the only one here, other than you?

presently, yes

that might change

ok. may I ask a question from our dev forums?

ok :)

sure

if you want, just paste in a link

It's on Contacts. Is the issue on our side, or his?

well, I haven't tried ACTION_PICK with a content type instead of a Uri

didn't know that worked

in fact, the more I ponder, I'm a bit surprised it does

yes, I wasn't sure either

it's on the data level, right?

not sure I understand that question

In 2.0+, Contacts has 3 levels ... data, raw, and contact

at least, according to the Android docs

oh, yeah, well, you just blew past my limited comprehension of the new Contacts system

the old system was bad enough

the new one gives me a headache just thinking about it

I know :-)

Me too :-)

I think what he's trying to do is display only contacts with a phone number

Is there a new/old way that works?

via ACTION_PICK, not sure

if you have the READ_CONTACTS permission, then definitely

So maybe that's our suggestion to him ...

the Contacts/Spinners sample out of the Advanced Android book demonstrates it, for old and new APIs

right, and I ran it on a DROID X, and it was fine

the real question is: is what he's trying to do actually supposed to work?

My guess is that our Blur syncs are changing IDs, and that's why he's getting a null back

his phrasing suggests that he has an app on the Market, and that it's working for other devices

I haven't seen it in any docs, articles, or samples

Yes, but developers tell us that all the time, to get us to debug their apps

oh, that's likely

"My app works on all devices except Motorola X .. "

I mean, who do they think I am? You?

heh

yeah

the last entry on there suggests a private message to you

did he send you an APK?

Yes, he sent me a pm, but no apk yet

Most I can do is run his APK on two devices and if needed, file a bug

But our Contacts has gone into maintenance ..

so it's unlikely to be fixed

so I'm looking for a coding technique for him

ah

here, somebody is using an Intent with that MIME type

but they also set a Uri

oh, so it is okay

right !!

you're a genius ..

so I'll refer him to that

here's another reference to the same concept: http://groups.google.com/group/android-dev…

ok, thanks

I just Googled on setType(ContactsContract.CommonDataKinds.Phone.CONTENT_TYPE);

those links and others showed up

ok

I'll google that too and see what else is there

I think we should just refer these devs directly to you

oy

To buy WareScriptions and use your chats. Yes?

oh?

sorry, I thought you meant direct emails

no, I want them to pay you :-)

Motorola is certainly welcome to refer people my way, and I certainly would appreciate such referrals

if I were in Motorola's shoes...well, I'd want to ponder that approach

why so?

I suppose it's better than just having threads tail off and moulder

Well, our direction is not to do 1-1 consulting, because our support is free

understood

perhaps it's just a matter of messaging

yes

So I'll refer him to those links, thanks so very much

no problem

And I'll mention to Lisa that we might to message what we do/don't do on forums

And/or refer devs to you, for WareScriptions, not direct email

yeah

I once spent 12+ hours debugging some guy's widget ..

He claimed it worked on all devices but DROID 2

Man, what a lesson

the key really is getting the developers to demonstrate that the problem exists on some MOTODEVICE but works fine on others, and that would typically require an APK

Right

Only thing is, if they do send it, and we do reproduce ..

We can't always get a fix anyway

true, but it gets back to messaging

Yes, entirely right

if the problem truly exists on a MOTODEVICE, then you're in normal MOTODEV support territory

if the problem exists on other hardware, you can steer them to StackOverflow or me or [android-developers] or wherever

Right

and, if they do not supply an APK, that's their own fault

I'm leaning how to handle these

Yes, I agree

it's why I always recommend attaching a project to an issue on b.android.com

same sort of thing applies here

b.android.com? bug site?

yes

issues seem to get incrementally more love if they are replicable

yes, that's the way of software engineering

alas, "software engineering" is a foreign concept to many developers

yes, especially those who build consumer apps, right?

for me, it seems more a question of background and scope

firms tend to be pretty decent

right, but individuals?

that depends on where they've worked and how much they internalized the process

or, whether they had to respond to public issue trackers... :-)

right

on the whole, at least in Android, a lot of the solo developers are pretty new to all of this

So the second link you sent

howdy, Sean!

Doesn't have a URI? Is that recommended?

Hi Mark!

Suzanne: oh, you're right

never mind that

sorry

Ok. You've been a great help

Have a wonderful holiday, talk soon

Sean: do you have a question?

you too!

Yes, I do, just a sec

It's about what the best practice would be to accomplish a particular task in Android.

um, ok

got a task in mind?

or can I pick one? :-)

I'm starting an app which will access information on my car's computer

It will have a lot of different functions

let's take one, a map for example.

I may display the car's location on the map on the phone.

I have an icon on the map representing the car.

I want to click on the icon and let the user choose from many Tasks (to use an Android term I've found in the documentation)

I really doubt you want to choose from Android tasks, but go on

so, each item in the list that the icon pops up will let the user perform a specific Task

for example, tasks might be, "find nearest" or "unlock doors" or whatever

but I decide to release this app and only some of the tasks are done.

I am thinking of writing each "task" as a separate application, with no icon

(btw, howdy, brack!)

An intent will launch the Activity for that "task"

(hello again, Mark!)

Sean: so far, that all sounds reasonable

So when the user starts the app, the app that shows the map goes to the web and discovers that "unlock doors" is now complete and ready as an app

and at this point I'd love to have my app start a download in the background and install that app but I know that is not possible yet.

well, you can do the download and kick off an install, but it won't install "quietly"

the user will still go through the permission check and such

Right, and it annoys the user.

:: shrug ::

only if they are not expecting it

that's a question of messaging

so I would wait until the user clicks the icon and I would display, "unlock doors"

but when the user clicks that, I would say, "hey, this feature is available, but will take a few clicks to install, wanna do that now?"

basically, what I'm trying to do is extend my app as features become available, rather than constantly bug a user to install a new version, for a feature they might not even care about.

that sounds great

Where is Android as far as making this kind of use case possible/?

in fact, in 2011, if nobody else solves it, I have plans on tackling this very problem

well, it's mostly there

you need to call out to a Web service or download a page or something to determine which of your plugins are ready

that will drive your list which has the "unlock doors" option

I think we're talking about a module system. hate to reinvent something.

sorta

these are all really standalone apps

they're just not designed for independent distribution on the Android Market

by "standalone", I mean they are APK files

there is no concept for a "plug-in" for android yet, is there? all have to be standalone?

well, the easy path is standalone

there are some hooks in there to download what amount to JARs of Dalvik bytecode

haven't tried them

I assume they are in the proprietary API

no

really!

dalvik.system.DexClassLoader

again, though, I haven't tried it

your kilometerage may vary

I will have to look at that one and see what restrictions there are

yeah, classloaders are tricky, even in regular Java

I can't imagine it is somehow easier on Android

anyway, let me take a question from brack, and we can swing back to you for followup if needed

brack: do you have a question?

and the other option would be to have that feature sit there on the Android market as an application that one can purchase. The problem is that as a "plug-in" it would not have a visible icon; if people bought it I suppose the first thing it would do when run (if the user figured out how to start it) would be to tell them to download the rest of the app

OK, go ahead. I'll stand by

yep, quick one, hopefully

right now I have a service that uses a MediaPlayer. the big issue right now is that when i scroll, say, a listview, it lags the audio very badly

i'm assuming the service isn't getting enough cpu resource

that's very possible

are you using startForeground()?

that's in the service, right?

yes

i don't know if i actually added that, no.

damn, i looked at that too. you think that'll do it?

I *think* that might keep your service in foreground priority

wasn't sure if it was that or threading, or if i needed to create a remote service and run it in a separate process

hopefully neither of those

yeah heh

try startForeground()

thanks- will lurk for a bit while i try this. appreciate it

no problem

Sean: there's a middle ground between DexClassLoader and APK-on-the-Market

you can always download and kick off the installs of APKs yourself

this still puts an icon in their Settings > Applications > Manage Applications screen

I'm OK with the icon in settings, just not in the main app list

just don't create an Activity with a LAUNCHER <intent-filter>

when you say "you can download" do you mean "you" as the user or "you" as the application?

"you" as the application

How would my app do that?

ummm...HTTP GET, presumably

because that sounds good enough. my app will look to see if the plug-in is installed, and if not, install it

then an ACTION_VIEW Intent with the APK MIME type

you could do an ACTION_VIEW on the APK URL directly, but I think the user needs to do an additional step as part of the install

That sounds very promising, I will give that a shot

you will have to deal with updates yourself

brack: do you have another question?

yeah, i was generally curious about the seekbar and if you can disable trackpad interaction

usually, no

either it is enabled (touch plus d-pad/trackball/whatever) or it is not

roger. main issue is that it's hooked up to a media player, and the movements with the trackpad seem to lag it out

you could subclass it and try to eat the events

nah, i'll just put a limiter on the callback for adjusting the seek position

thanks, that answers that one. appreciate it. how long are you online? till 5?

yes

Sean: got another question?

yes

let's say I use the "plug-in" approach to sell a feature at the point it is needed. The user is directed to the market to install the plug-in. I see several problems with this: (1) the plugin has no LAUNCHER, so if someone buys it wituout the app they will not be happy and (2) how to return to the app once the plug-in is installed

(2) is the user pressing BACK

(1) is indubitably a problem

which is why I am hoping we in the community can work out a plugin mechanism, with an update manager, for this sort of thing

both significant user experience problems

I am looking for a project just like that!

where is this being discussed?

I guess I mean, is the community already discussing this, and if so, where?

by the voices inside my head

:-)

it's an itch I will need to scratch at some point, if nobody else scratches it

I am not aware of anyone working on something quite like this, though it's a big community

that actually sounds really interesting, a plugin updating manager (if I've been reading correctly)

Well, I will put it on my own short list of things to research.

yes

I believe OSGi is meant to do this

my personal itch will be for more of a general non-Market-APK updating manager, but it could be applied for plugins as well

I know some people have been poking at Apache Felix vis-a-vis Android

however, OSGi feels like engineering-by-committee

right, that is my concern too

it would need to be something that can be done in pieces or it will be too big and never get done

OK, our time is running out -- any last questions?

one quick one if you have a sec: does it matter where i call startforeground?

anywhere in the service OK, or does it have to be in startCommand

sometime before you need foreground priority

gotcha; great, thanks!

since it puts an icon in the launcher, I'd do it at the point the user did whatever it was that caused your service to need foreground priority

that way, user sees cause-and-effect

versus an icon appearing out of proverbial nowhere

er, not icon in the launcher

icon in the *taskbar*

wow, I need sleep

right- yeah. that's what i'm doing now

haha

if brack has no others I might ask: I'm on the Java ME executive committee for the JCP, watching the SE/EE vote on SE7/SE8; do you have any opinions on Java's future with Oracle? anywhere you'd prefer it to go?

i was curious because it didn't help, but i'm wondering if i need to call that out before I do the media player initialization, etc

but that's all for now, I may check in for tomorrow's hours once i play with threading and separate processes

brack: sounds good!

Sean: my arms are too short to box with Oracle

thanks again. cya later guys

thanks! bye!

a great answer to my last question. I need to realize that myself. take care.

see ya!