Hi Mark

hello, Andrei!

how can I help you today?

I have a few questions, all revolving around a context with multiple fragments

I have a LinearLayout with two fragments, one with a list of items and the other one with details about the item currently selected in the first fragment. The first fragment is embedded statically in the layout, whereas the other one dynamically via a FrameLayout. Each fragment has its own loader as they interrogate different tables. When adding another item, I launch a third fragment which replaces the LinearLayout. When finishing adding a new item, I do a "getFragmentManager().popBackStack()" to get back to the previous layout with two fragments. However, the new item does not appear in the item list from the first fragment. I know the loaders do some caching. If this is the problem, how can I force a refresh?

"When adding another item" = "When adding another item to the 1st framgent's list of items"

how are you "adding a new item"? are you calling insert() on a ContentResolver or something?

I call insert on a dbHelper, I don't need a ContentResolver

ummm... what sort of Loader are you using, then?

your loaderex :-)

oh

I use SqlCipher

then you should be calling insert() on the loader

so the loader knows to reload its Cursor for the UI

that means that I should return the data to the 1st fragment and that one should use the loader to insert the data

not necessarily

I don't have access to the loader in the 3rd fragment currently

and that's one of the reasons why I have discontinued this library

in short, loaders suck *except* for CursorLoader and a ContentProvider

so, if you wish to continue using SQLCipherCursorLoader, you would need to either get the Loader someplace where all fragments in the activity can reach it, or you would need to pass the data to be inserted to the first fragment, so it can do the insert

Ok, I'll read that. I know you wanted to discontinue that. I imagine it's also a headache to answer all people who use loaderex...

it's more that it just doesn't work very well in practice

so I never use it

so I don't maintain it

if it were something that "just worked", that'd be one thing

ok.

hi all

but, as I mentioned, it sucks, and so not maintaining it will only make it suck worse over time

hello, Luca!

Luca: it's your turn -- do you have a question?

Hi Luca

yes

i'm using some custom views and to avoid too many allocations i'm caching objects like rect and paint in static variables, obviously only those that i know don't change when using the same view multiple times, is it a good practice?

"i'm caching objects like rect and paint in static variables" -- yuck!

caching them as data members may be OK, done carefully

caching them as statics seems like overkill and risky

My turn?

Luca: I'll swing back to you in a bit to follow up

Andrei: go ahead

Given the same context as before: initially I was also forcing an "artificial click" on an item from the first fragment's list via some code placed in "onLoadFinished()", in order to populate the details fragment. Now after adding the 3rd fragment to add items to the list in the 1st fragment, my program crashes when getting back from the 3rd fragment to the first two fragments. Namely, "getlistview()" which I used in "onLoadFinished()" crashes with the message "java.lang.IllegalStateException Content view not yet created". How do I force a selection without getting into this trouble?

setSelection() isn't working for you?

though, reading the JavaDocs, that might not work in touch mode...

is your ListView single-choice mode?

I think it didn't work, if I recall correctly

yes

then it'd be setItemChecked()

which should toggle your activated state, assuming that you are using one

ok, I'll try that

When people had the same error it seemed that the loader got activated before the view was re-created

or something like that

Luca: do you have another question, or a follow-up from the first one?

(but not with your loader)

Luca: if you do come up with another question, say so, and I'll route back to you

Andrei: if you have more, go ahead

One more from me: I notice that when I come back from the 3rd fragment, my 1st fragment doesn't go through "onCreateView()", though the Fragments documentation suggests it should: http://developer.android.com/images/fragment_li... . Is this normal?

your first fragment would not have gone through onPause()/onStop()/onDestroyView() just by being pushed onto the back stack

hence, there would be no onCreateView() when coming back after popBackStack()

your view is already created

But it goes through the others after onCreateView(), when returning from the 3rd fragment

what are "the others"?

I did not check onPause and the following, but I checked everything from onCreateView() on

i.e. onActivityCreated(), onStart() and onResume()

:: shrug :: I don't know what to tell you, sorry

i haven't other questions, i will avoid to use static rect/paint in views maybe it was an overoptimization/overkill furthermore views are already cached by adapters, i'm trying to allocate fewer objects possible to reduce the intervention of the garbage collector because i'm afraid of it.. "GC stops the world"

it goes through all these but not through onCreateView() so I was wondering why, but it's not an essential question anyway

coming back to your answer to my first question, you suggested to make the loaders available globally to all fragments. This means I should use some static variables to hold them?

Luca: GC doesn't usually "stop the world" on Android 3.0+, though it can consume CPU cycles

Andrei: no, have the Activity hold it, and the fragments can talk to the activity

assuming the 3rd fragment was the one doing the insert(), that's all the broader access you'd need

indeed

they all belong to the same activity

that was all from me

thanks for your time and advices!

you are very welcome

actually, could one or both of you do me a quick small favor, and try visiting https://stackoverflow.com in a browser?

I'm seeing an untrusted certificate error, and I'm trying to determine if that's just my environment

I get a "This is probably not the site that you are looking for!"

SSL Error, indeed

they're probably doing work related to the "heartbleed" OpenSSL bug

i get the error too

thanks!

that's a wrap for today's chat

the next one is Monday at 4pm US Eastern Daylight Time

the chat transcript will be posted shortly to http://commonsware.com/office-hours/

have a pleasant day!