Develop Reference

When we should use fragment and when we should use activity? [duplicate]

I know that Activities are designed to represent a single screen of my application, while Fragments are designed to be reusable UI layouts with logic embedded inside of them.
Until not long ago, I developed an application as it said that they should be developed.
I created an Activity to represent a screen of my application and used Fragments for ViewPager or Google Maps. I rarely created a ListFragment or other UI that can be reused several times.
Recently I stumbled on a project that contains only 2 Activities one is a SettingsActivity and other one is the MainActivity. The layout of the MainActivity is populated with many hidden full screen UI fragments and only one is shown. In the Activity logic there are many FragmentTransitions between the different screens of the application.
What I like about this approach is that because the application uses an ActionBar, it stays intact and does not move with the screen switching animation, which is what happens with Activity switching. This give a more fluent feel to those screen transitions.
So I guess what I'm asking is to share your current development manner regarding this topic, I know it might look like an opinion based question at first look but I look at it as an Android design and architecture question... Not really an opinion based one.
UPDATE (01.05.2014): Following this presentation by Eric Burke from Square, (which I have to say is a great presentation with a lot of useful tools for android developers. And I am not related in any way to Square)
http://www.infoq.com/presentations/Android-Design/
From my personal experience over the past few months, I found that the best way to construct my applications is to create groups of fragments that come to represent a flow in the application and present all those fragments in one Activity. So basically you will have the same number of Activities in your application as the number of flows.
That way the action bar stays intact on all the flow's screens, but is being recreated on changing a flow which makes a lot of sense. As Eric Burke states and as I have come to realize as well, the philosophy of using as few Activities as possible is not applicable for all situations because it creates a mess in what he calls the "God" activity.

Experts will tell you: "When I see the UI, I will know whether to use an Activity or a Fragment". In the beginning this will not have any sense, but in time, you will actually be able to tell if you need Fragment or not.
There is a good practice I found very helpful for me. It occurred to me while I was trying to explain something to my daughter.
Namely, imagine a box which represents a screen. Can you load another screen in this box? If you use a new box, will you have to copy multiple items from the 1st box? If the answer is Yes, then you should use Fragments, because the root Activity can hold all duplicated elements to save you time in creating them, and you can simply replace parts of the box.
But don't forget that you always need a box container (Activity) or your parts will be dispersed. So one box with parts inside.
Take care not to misuse the box. Android UX experts advise (you can find them on YouTube) when we should explicitly load another Activity, instead to use a Fragment (like when we deal with the Navigation Drawer which has categories). Once you feel comfortable with Fragments, you can watch all their videos. Even more they are mandatory material.
Can you right now look at your UI and figure out if you need an Activity or a Fragment? Did you get a new perspective? I think you did.

My philosophy is this:
Create an activity only if it's absolutely absolutely required. With the back stack made available for committing bunch of fragment transactions, I try to create as few activities in my app as possible. Also, communicating between various fragments is much easier than sending data back and forth between activities.
Activity transitions are expensive, right? At least I believe so - since the old activity has to be destroyed/paused/stopped, pushed onto the stack, and then the new activity has to be created/started/resumed.
It's just my philosophy since fragments were introduced.

Well, according to Google's lectures (maybe here, I don't remember) , you should consider using Fragments whenever it's possible, as it makes your code easier to maintain and control.
However, I think that on some cases it can get too complex, as the activity that hosts the fragments need to navigate/communicate between them.
I think you should decide by yourself what's best for you. It's usually not that hard to convert an activity to a fragment and vice versa.
I've created a post about this dillema here, if you wish to read some further.

Since Jetpack, Single-Activity app is the preferred architecture. Usefull especially with the Navigation Architecture Component.
source

Why I prefer Fragment over Activity in ALL CASES.
Activity is expensive. In Fragment, views and property states are separated - whenever a fragment is in backstack, its views will be destroyed. So you can stack much more Fragments than Activity.
Backstack manipulation. With FragmentManager, it's easy to clear all the Fragments, insert more than on Fragments and etcs. But for Activity, it will be a nightmare to manipulate those stuff.
A much predictable lifecycle. As long as the host Activity is not recycled. the Fragments in the backstack will not be recycled. So it's possible to use FragmentManager::getFragments() to find specific Fragment (not encouraged).

In my opinion it's not really relevant. The key factor to consider is
how often are you gonna reuse parts of the UI (menus for example),
is the app also for tablets?
The main use of fragments is to build multipane activities, which makes it perfect for Tablet/Phone responsive apps.

Don't forget that an activity is application's block/component which can be shared and started through Intent! So each activity in your application should solve only one kind of task. If you have only one task in your application then I think you need only one activity and many fragments if needed. Of course you can reuse fragments in future activities which solve another tasks. This approach will be clear and logical separation of tasks. And you no need to maintain one activity with different intent filter parameters for different sets of fragments. You define tasks at the design stage of the development process based on requirements.

There's more to this than you realize, you have to remember than an activity that is launched does not implicitly destroy the calling activity. Sure, you can set it up such that your user clicks a button to go to a page, you start that page's activity and destroy the current one. This causes a lot of overhead. The best guide I can give you is:
** Start a new activity only if it makes sense to have the main activity and this one open at the same time (think of multiple windows).
A great example of when it makes sense to have multiple activities is Google Drive. The main activity provides a file explorer. When a file is opened, a new activity is launched to view that file. You can press the recent apps button which will allow you to go back to the browser without closing the opened document, then perhaps even open another document in parallel to the first.

Thing I did: Using less fragment when possible. Unfortunately, it's possible in almost case. So, I end up with a lot of fragments and a little of activities.
Some drawbacks I've realized:
ActionBar & Menu: When 2 fragment has different title, menu, that
will hard to handle. Ex: when adding new fragment, you can change action bar title, but when pop it from backstack there is no way to restore the old title. You may need an Toolbar in every fragment for this case, but let believe me, that will spend you more time.
When we need startForResult, activity has but fragment hasn't.
Don't have transition animation by default
My solution for this is using an Activity to wrap a fragment inside. So we have separate action bar, menu, startActivityForResult, animation,...

The one big advantage of a fragment over activity is that , the code which is used for fragment can be used for different activities. So, it provides re-usability of code in application development.

use one activity per application to provide base for fragment
use fragment for screen ,
fragments are lite weight as compared to activites
fragments are reusable
fragments are better suited for app which support both phone & tablet

You are free to use one of those.
Basically, you have to evaluate which is the best one to your app. Think about how you will manage the business flow and how to store/manage data preferences.
Think about, how Fragments store garbage data. When you implement the fragment, you have a activity root to fill with fragment(s). So, if your trying to implement a lot of activities with too much fragments, you have to consider performance on your app, coz you're manipulating (coarsely speaks) two context lifecycle, remember the complexity.
Remember: should I use fragments? Why shouldn't I?
regards.

I use Fragments for better user experience. For example if you have a Button and you want to run let's say a webservice when you click it, I attach a Fragment to the parent Activity.
if (id == R.id.forecast) {
ForecastFragment forecastFragment = new ForecastFragment();
FragmentManager fm = getSupportFragmentManager();
FragmentTransaction ft = fm.beginTransaction();
ft.replace(R.id.main_content, forecastFragment);
ft.addToBackStack("backstack");
forecastFragment.setArguments(b);
ft.commit();
}
In that way the user won't have to move in another activity.
And secondly I prefer Fragments because you can handle them easily during rotation.

It depends what you want to build really. For example the navigation drawer uses fragments. Tabs use fragments as well. Another good implementation,is where you have a listview. When you rotate the phone and click a row the activity is shown in the remaining half of the screen. Personally,I use fragments and fragment dialogs,as it is more professional. Plus they are handled easier in rotation.

Almost always use fragments. If you know that the app you are building will remain very small, the extra effort of using fragments may not be worth it, so they can be left out. For larger apps, the complexity introduced is offset by the flexibility fragments provide, making it easier to justify having them in the project.
Some people are very opposed to the additional complexity involved with fragments and their lifecycles, so they never use them in their projects. An issue with this approach is that there are several APIs in Android that rely on fragments, such as ViewPager and the Jetpack Navigation library. If you need to use these options in your app, then you must use fragments to get their benefits.
Excerpt From: Kristin Marsicano. “Android Programming: The Big Nerd Ranch Guide, 4th Edition.” Apple Books.

Some wrong ideas:
Always put an activity in your app and handle different screens with fragments.
Write the UI code directly in the activity.
Handle navigating between screens (I don't mean tabs, I mean for example full-screen views) by fragments.
Activities can be replaced by fragments.
Here is the thing!
Fragments are designed to implement reusable parts of UI and use them in any part of the app that is needed. They are not designed for substituting activities.
When we must use each of them?
When we have an independent screen in which there are some different UI parts (tabs, expandable screens, partial screens, etc...) we should use an activity with some fragments to implement and handle different UI parts separately in the same screen.
Each independent part of the application is actually a component that is conceptually different from other parts and it needs to have an independent activity. For example, the login part may contain some different scenarios like using username-password or using fingerprint. Each scenario can be implemented by a fragment and all login-related fragments should be handled by LoginActivity. But for example, the part of the orders in the application doesn't have a conceptual relationship with login, so it must have a different activity and of course, it may contain some fragments like OrdersFragment, SubmitNewOrderFragment, and etc that all of them must be managed by OrdersActivity.
Do not implement a UI directly in an activity. Always implement UI in fragments and add those fragments in the activity even if there is only one fragment in that activity. It helps you to have more reusable code and change UIs easier.
Never use fragments to navigate infinitely in the application even if you force the user to have a limited number of fragments in the back stack. The fact is that when you add a new fragment into the back stack and remove it, it will not be removed from the memory unless the parent activity gets destroyed and it is only not visible. So when you are using fragment manager back stack, by navigating multiple times between fragments in the same activity (especially in the case of you create a new fragment on each navigation and put them into the back stack) you will get an OutOfMemoryException in the application.
I hope it to be helpful.

This question needs to be reevaluated since Jetpack Compose has reached stable.
Jetpack Compose is Android’s recommended modern toolkit for building
native UI.
from https://developer.android.com/jetpack/compose
The typical jetpack-compose architecture is:
Single Activity, multiple composables, and glued together by jetpack navigation.
Note there's no (need for) Fragments anymore.
See Now in Android for a sample.

When to use fragment backstack vs. individual activities?

I've been developing for Android for a couple of years now, and I still find myself going back and forth on this issue: when should I be using fragments on a backstack versus putting each fragment into its own activity?
In the Android Fragment documentation, they show this diagram:
I understand the tablet use case perfectly, but for the Handset use case, I don't get why you would put each fragment into its own activity. I typically create one activity and add the fragments to the backstack (via FragmentManager). Is either one of these approaches considered the 'right' way to do it? If both are okay, what is a good rule of thumb to use for picking which approach to use?
This question is closely related, but I'm not totally satisfied with it. If you're supposed to use separate activities, what's the point of having a fragment backstack in the first place?

I doubt there is a right way to do it. There are likely some ways that are better than others in some circumstances, but for the most part, the answer will be "it depends."
I have noticed in the latest version (22.x) of the SDK with Eclipse that every Activity that is generated is nothing more than a placeholder for a Fragment. The Fragment is auto generated and contains the view logic. It seems like they want to make Activities nothing more than a placeholder/controller for Fragments. I don't think I agree with that. While I definitely see the use case for Tablets, I feel like this pattern should be used more on an as needed basis than as a general rule. This is just my opinion, but I think moving ALL of the logic into the Fragment sacrifices some of the benefits you get from using an Activity, so it is only a useful pattern if you need to re-use that fragment specifically.
If you're supposed to use separate activities, what's the point of having a fragment backstack in the first place?
Good question. I personally prefer to use many Activities and only use Fragments where I need to re-use the logic/views in multiple places. I think the flow of using startActivity and startActivityForResult and allowing the system to manage your activity stack is just a little easier than trying to manage a huge Fragment back stack and one Activity (again, just my opinion).
So when would I ever use the Fragment back stack? There was actually a very good situation for me to use it recently. I had an Activity that needed to build a very complex object. The Object needed many fields input to the user, so I created a Workflow that walked the user through this process one step at a time. I created a single Activity to handle the creation of this logic. Each step of the UI was a Fragment that took input from the user, reported back to the Activity, then the Activity loaded the next Fragment. The Fragments were added to the back stack so the user could go back to previous steps in the workflow.
Object1CreationActivity
FragmentA --> FragmentB --> FragmentC --> FragmentD
Communication between the Fragments and the Activity should be done with interfaces. This is important if we want to re-use these anywhere else. Because of this, I could re-use much of this code to create another Object.
Object2CreationActivity
FragmentB --> FragmentD --> FragmentE
To summarize, Fragments, Activities, the back stack, these are all powerful tools you can use to make Android applications. There may not be a great rule of thumb for when and how to use them, but as long as you are well versed in how they work together, you can use them as appropriate for your application.

fragments are light weight alternatives to activities....that is one way to look at it. e.g. i have my our app which has about 10-13 screens. Either
I create a new activity for each one of them. OR
I create only a few activities that are logically distinct in functional aspects ad swap view screens in them OR
I create 1 activity and delegate actual screen functional to fragments.
I find the third way much better and manageable. Its kinda saying that fragments allow reuse of view using the framelayout options. Further more you can devise an easy way to share data between fragments as against the heavy weight way of using a Parceable to share stuff between activities.
Also Android API folks are going to focus more on fragment based designs instead of activities so its better to stick with standards. Using fragments i a just a tad more complicated than using activities but is well worth the effort to learn them. Helps create scalable and quiet reusable screen views IMHO.

One Activity and all other Fragments [closed]

Closed. This question is opinion-based. It is not currently accepting answers.
Want to improve this question? Update the question so it can be answered with facts and citations by editing this post.
Closed 9 years ago.
Improve this question
I am thinking of implementing one screen with Activity and all other sreens with Fragments and managing all the fragments thru the activity.
Is it a good idea? and my answer is NO but still I want to know more clearly about this thought.
What are the pros and cons of the idea?
Note:
Please don't give me the link for fragment and activity.
EDIT:
Here is something over Fragments and activity:
Pros:
Fragments are meant to be used with activities as a sub activity.
Fragments are not the replacement for activities.
Fragments are meant for reusability(Need to know in what way reusability can be achieved.).
Fragments are the best way to write code to support both tablets and phones.
Cons:
We need to implement the interface to get the data from fragments.
For dialog we have to go a long way to show it.
Why should we use fragments if we are not considering tablets?
What is the starting time difference between activity and fragment?

It depends on the app you are creating. I've created several apps using both approaches and can't say one way is always better than the other. The latest app I created I used the single Activity approach and a Facebook style navigation. When selecting items from the navigation list I update a single Fragment container to display that section.
That said, having a single Activity also introduces a lot of complexities. Let's say you have an edit form, and for some of the items the user needs to select, or create, requires them to go to a new screen. With activities we'd just call the new screen with startActivityForResult but with Fragments there is no such thing so you end up storing the value on the Activity and having the main edit fragment check the Activity to see if data has been selected and should be displayed to the user.
What Aravind says about being stuck to a single Activity type is also true but not really that limiting. Your activity would be a FragmentActivity and as long as you don't need a MapView then there are no real limitations. If you do want to display maps though, it can be done, but you'll need to either modify the Android Compatibility Library to have FragmentActivity extend MapActivity or use the the publicly available android-support-v4-googlemaps.
Ultimately most the devs I know that went the one Activity route have gone back to multiple Activities to simplify their code. UI wise, on a tablet, you are some times stuck using a single Activity just to achieve what ever crazy interaction your designers come up with :)
-- EDIT --
Google has finally released MapFragment to the compatibility library so you no longer have to use the android-support-v4-googlemaps hack. Read about the update here: Google Maps Android API v2
-- EDIT 2 --
I just read this great post about the modern (2017) state of fragments and remembered this old answer. Thought I would share: Fragments: The Solution to All of Android's Problems

I'm about to finish a project(5 months in development), that has 1 activity, and 17 fragments, all full screen. This is my second fragment based project(previous was 4 months).
Pros
The main activity is 700 lines of code, just nicely managing the order of the fragments navigation.
Each fragment is nicely separated into it's own class, and is relatively small (~couple hundred lines of ui stuff).
Management can say, "hey, how about we switch the order of those screens", and I can do it very easily, as those fragments don't depend on each other, they all communicate through the activity. I don't have to dig through individual activities, to find where they call each other.
my app is very graphics heavy, and would never work as 1 screen 1 activity. All those sub activities in the memory, would make the app run out of memory all the time, so I would have to finish() all non visible activities, and make the same control logic for navigation, as I would do with fragments. Might as well do it with fragments just because of this.
if we ever do a tablet app, we will have an easier time re-factoring stuff, because everything is nicely separated already.
Cons
you have to learn, how to use fragments

First, whatever you do, make sure you have a modular design using model, view, presenter that is not highly dependent on an Activity or a Fragment.
What do Activities and Fragments really provide?
Life cycle events and backstack
Context and resources
Therefore, use them for that, ONLY. They have enough responsibility, don't over complicate them. I would argue that even intantiating a TextView in an Activity or Fragment is bad practice. There is a reason methods like public View findViewById (int id) are PUBLIC.
Now the question gets simpler: Do I need multiple, independent life cycle events and backstacks? If you think yeah maybe, use fragments. If you think never ever, don't use fragments.
In the end, you could make your own backstack and life cycles. But, why recreate the wheel?

Pros
You could control your fragments from a single activity, beacause all fragments are independent of each other. The fragments have a lifecycle (onPause, onCreate, onStart...) of their own. By having a lifecycle, fragments can respond independently to events, save their state through onSaveInstanceState, and be brought back (i.e. such as when resuming after an incoming call or when the user clicks the back button).
Cons
Create complexity in your code of activity.
You have to manage the order of the fragments.
Never the less, it is quite a good idea, as if you need to create an app, where you want to show several views. By this idea, you'll be able to view several fragments in a single view..

It depends on the design layout of your app. Suppose if your using Tabs in ActionBar in the design layout then in the Single Activity of the app one can have fragments being changed on Tab click. So now you have an Activity and say suppose three Tabs in the ActionBar and the view for the tabs being provided by the Fragments, which makes it easy to manage plus is feasible also.
So, it all depends on the design scheme of your app and how you take the decision to build for it.

Pros:
Can be used to create a single interface usable by multiple screen sizes and orientations via xml layouts.
Cons:
Requires more complex code in your activity.
I believe it's a good idea, because using different xml layouts based on the current screen size and orientation can make the app more usable and reduce the need to release multiple versions of your app if you plan on releasing your app for both phones and tablets. If your app will never be used by both tablets and phones, it's probably not worth the trouble.

I'm a proponent of deferring all view inflation to fragments to provide better flexibility. For example having a single landing activity for a tablet which aggregates multiple fragments and reusing the same fragments on a phone to show one screen per fragment. However in the phone implementation I'd have a separate activity for each screen. The activities would not have too much code as they would immediately defer to their fragment counterpart for view inflation.
I think it's a bad idea for the phone implementation to have to change to a single landing activity when tabs or a slide out menu is introduced since the tab or menu navigation just results in a completely new screen.

The most important reason I would give for not using the single activity approach is so that the activity lifecycle can be taken advantage of. Activities contain contextual behavior of a certain portion of the application and fragments supplement that behavior. Having the ability to take advantage of the overridable steps in the activity lifecycle helps to separate one activity's behavior from another with methods such as onPause and onResume. This lifecycle also allows you return to previous context. With the single activity approach, once you leave a fragment you have to create an mechanism to return to it.

Why fragments, and when to use fragments instead of activities?

In Android API 11+, Google has released a new class called Fragment.
In the videos, Google suggests that whenever possible (link1, link2), we should use fragments instead of activities, but they didn't explain exactly why.
What's the purpose of fragments and some possible uses of them (other than some UI examples that can be easily be achieved by simple views/layouts)?
My question is about fragments:
What are the purposes of using a fragment?
What are the advantages and disadvantages of using fragments compared to using activities/views/layouts?
Bonus questions:
Can you give some really interesting uses for fragments? Things that Google didn't mention in their videos?
What's the best way to communicate between fragments and the activities that contain them?
What are the most important things to remember when you use fragments? Any tips and warnings from your experience?

#1 & #2 what are the purposes of using a fragment & what are the
advantages and disadvantages of using fragments compared to using
activities/views/layouts?
Fragments are Android's solution to creating reusable user interfaces. You can achieve some of the same things using activities and layouts (for example by using includes). However; fragments are wired in to the Android API, from HoneyComb, and up. Let me elaborate;
The ActionBar. If you want tabs up there to navigate your app, you quickly see that ActionBar.TabListener interface gives you a FragmentTransaction as an input argument to the onTabSelected method. You could probably ignore this, and do something else and clever, but you'd be working against the API, not with it.
The FragmentManager handles «back» for you in a very clever way. Back does not mean back to the last activity, like for regular activities. It means back to the previous fragment state.
You can use the cool ViewPager with a FragmentPagerAdapter to create swipe interfaces. The FragmentPagerAdapter code is much cleaner than a regular adapter, and it controls instantiations of the individual fragments.
Your life will be a lot easier if you use Fragments when you try to create applications for both phones and tablets. Since the fragments are so tied in with the Honeycomb+ APIs, you will want to use them on phones as well to reuse code. That's where the compatibility library comes in handy.
You even could and should use fragments for apps meant for phones only. If you have portability in mind. I use ActionBarSherlock and the compatibility libraries to create "ICS looking" apps, that look the same all the way back to version 1.6. You get the latest features like the ActionBar, with tabs, overflow, split action bar, viewpager etc.
Bonus 2
The best way to communicate between fragments are intents. When you press something in a Fragment you would typically call StartActivity() with data on it. The intent is passed on to all fragments of the activity you launch.

Not sure what video(s) you are referring to, but I doubt they are saying you should use fragments instead of activities, because they are not directly interchangeable. There is actually a fairly detailed entry in the Dev Guide, consider reading it for details.
In short, fragments live inside activities, and each activity can host many fragments. Like activities, they have a specific lifecycle, unlike activities, they are not top-level application components. Advantages of fragments include code reuse and modularity (e.g., using the same list view in many activities), including the ability to build multi-pane interfaces (mostly useful on tablets). The main disadvantage is (some) added complexity. You can generally achieve the same thing with (custom) views in a non-standard and less robust way.

A Fragment is a piece of an application's user interface or behavior that can be placed in an Activity which enable a more modular activity design. It will not be wrong if we say a fragment is a kind of subactivity.
Following are important points about a fragment:
A fragment has its own layout and its own behavior with its own lifecycle callbacks.
You can add or remove fragments in an activity while the activity is running.
You can combine multiple fragments in a single activity to build a multi-pane UI.
A fragment can be used in multiple activities.
The fragment life cycle is closely related to the lifecycle of its host activity.
When the activity is paused, all the fragments available in the acivity will also be stopped.
A fragment can implement a behavior that has no user interface component.
Fragments were added to the Android API in Android 3 (Honeycomb) with API version 11.
For more details, please visit the official site, Fragments.

Activities are the full screen components in the app with the toolbar, everything else are preferably Fragments.
One full screen parent activity with a toolbar can have multiple panes, scrollable pages, dialogs, etc. (all fragments), all of which can be accessed from the parent and communicate via the parent.
Example:
Activity A, Activity B, Activity C:
All activities need to have same code repeated, to show a basic
toolbar for example, or inherit from a parent activity (becomes
cumbersome to manage).
To move from one activity to the other, either all of them need to be in memory (overhead) or one needs to be destroyed for the other to open.
Communication between activities can be done via Intents.
vs
Activity A, Fragment 1, Fragment 2, Fragment 3:
No code repetition, all screens have toolbars etc. from that one activity.
Several ways to move from one fragment to next - view pager, multi pane etc.
Activity has most data, so minimal inter-fragment communication needed. If still necessary, can be done via interfaces easily.
Fragments do not need to be full screen, lots of flexibility in designing them.
Fragments do not need to inflate layout if views are not necessary.
Several activities can use the same fragment.

This is important information that I found on fragments:
Historically each screen in an Android app was implemented as a separate Activity. This creates a challenge in passing information between screens because the Android Intent mechanism does not allow passing a reference type (i.e. object) directly between Activities. Instead the object must be serialized or a globally accessible reference made available.
By making each screen a separate Fragment, this data passing headache
is completely avoided. Fragments always exist within the context of a
given Activity and can always access that Activity. By storing the
information of interest within the Activity, the Fragment for each
screen can simply access the object reference through the Activity.
Source: https://www.pluralsight.com/blog/software-development/android-fragments

Fragments are of particular use in some cases like where we want to keep a navigation drawer in all our pages. You can inflate a frame layout with whatever fragment you want and still have access to the navigation drawer.
If you had used an activity, you would have had to keep the drawer in all activities which makes for redundant code. This is one interesting use of a fragment.
I'm new to Android and still think a fragment is helpful this way.

I know this was already discussed to death, but I'd like to add some more points:
Frags can be used to populate Menus and can handle MenuItem clicks on their own. Thus giving futher modulation options for your Activities. You can do ContextualActionBar stuff and so on without your Activity knowing about it and can basically decouple it from the basic stuff your Activity handles (Navigation/Settings/About).
A parent Frag with child Frags can give you further options to modulize your components. E.g. you can easily swap Frags around, put new Frags inside a Pager or remove them, rearrange them. All without your Activity knowing anything about it just focusing on the higher level stuff.

Fragment can be thought of as non-root components in a composite tree of ui elements while activities sit at the top in the forest of composites(ui trees).
A rule of thumb on when not to use Fragment is when as a child the fragment has a conflicting attribute, e.g., it may be immersive or may be using a different style all together or has some other architectural / logical difference and doesn't fit in the existing tree homogeneously.
A rule of thumb on when to prefer Activity over Fragment is when the task (or set of coherent task) is fully independent and reusable and does some heavy weight lifting and should not be burdened further to conform to another parent-child composite (SRP violation, second responsibility would be to conform to the composite). For e.g., a MediaCaptureActivity that captures audio, video, photos etc and allows for edits, noise removal, annotations on photos etc and so on. This activity/module may have child fragments that do more granular work and conform to a common display theme.

Fragments lives within the Activity and has:
its own lifecycle
its own layout
its own child fragments and etc.
Think of Fragments as a sub activity of the main activity it belongs to, it cannot exist of its own and it can be called/reused again and again. Hope this helps :)

A fragment lives inside an activity, while an activity lives by itself.

If you've ever written front-end before, so used front-end components like(React, Vue or Angular). Think about fragments like reusable components within an activity.

1.Purposes of using a fragment?
Ans:
Dealing with device form-factor differences.
Passing information between app screens.
User interface organization.
Advanced UI metaphors.

Why fragments?
Fragments were created to replace most of the activity use cases. And see this Android Dev Summit session.
When to use fragments instead of activities?
Always, unless you really need an API that is only available in activity.

Two panel UI with Fragments vs Separate activities

I am starting a Honeycomb application that will have a basic two panel layout, one panel on the left for the menu and one on the right for the main functionality of each section.
Contrary to the available samples of the Fragments API the content displayed on the right panel consists of a completely different UI for each of the menu options.
It is tempting to just replace the right fragment according to the selected section, but this would mean using just one activity in the entire app, and this does not sound good. Moreover, the fragment's lifecycle is tied to the activity, so no fragments will be killed until the activity gets killed, resulting in a lot of fragments "alive".
However, having a different activity with two panels for every menu option means that the fragment used for the menu will have to be added in EVERY activity and will be subject to inconsistent layouts across all the sections that should have a menu.
What are the best practices here?

This blog post summarizes the reasons for choosing fragments over activities:
Embedded Activities via ActivityGroup
were a nice idea, but have always been
difficult to deal with since Activity
is designed to be an independent
self-contained component instead of
closely interacting with other
activities. The Fragment API is a much
better solution for this, and should
be considered as a replacement for
embedded activities.
Retaining data across Activity
instances could be accomplished
through
Activity.onRetainNonConfigurationInstance(),
but this is fairly klunky and
non-obvious. Fragment replaces that
mechanism by allowing you to retain an
entire Fragment instance just by
setting a flag.
A specialization of Fragment called
DialogFragment makes it easy to show a
Dialog that is managed as part of the
Activity lifecycle. This replaces
Activity’s “managed dialog” APIs.
Another specialization of Fragment
called ListFragment makes it easy to
show a list of data. This is similar
to the existing ListActivity (with a
few more features), but should reduce
the common question about how to show
a> list with some other data.
The information about all fragments
currently attached to an activity is
saved for you by the framework in the
activity’s saved instance state and
restored for you when it restarts.
This can greatly reduce the amount of
state save and restore code you need
to write yourself.
The framework has built-in support for
managing a back-stack of Fragment
objects, making it easy to provide
intra-activity Back button behavior
that integrates the existing activity
back stack. This state is also saved
and restored for you automatically.
Fragments are fairly new, so beyond that article, I'm not sure your going to find much for best practices. I think the decision you need to make is are my interactions tightly coupled and meant to share data or are they stand alone components which don't have much interaction.
edit, clarification: I think that using a single activity for an app isn't necessarily a bad decision. It's really a decision you should make based on the functionality of your app. Based on the article, an Activity is stand alone while a fragment is, typically, only relevant when combined with other fragments in the scope of an Activity. The situation you describe, with combinations of different activities is one of the pain points they designed Fragments to solve.