Share
Embed
Summoning a Zombie App
Slides for the Speech with the same name at BetabeersGRX, 2015/07/08.
Olmo Gallegos
July 08, 2015
Transcript
46 slides
-
<< A real story about targeting our efforts >> Summoning
a Zombie App Victor Olmo Gallegos Hernández Mobile developer -
MyTricks Notebook. The beginnings ◦ 2012 - Side project ◦
Satisfying my own needs ◦ Learning native development (Android SDK) -
About the App ◦ Personal notebook for Breakdance (bboying) ◦
Store your tricks by categories ◦ Add pictures, videos, text commentaries -
About the App - Beginnings ◦ Standalone App (not published)
◦ Very good feedback from friends ◦ A whole lot of motivation ◦ Helping everybody would be a good idea ◦ … Google Play submission - late 2012 -
Google Play submission ◦ Really good feedback ◦ Lots of
propositions (some of them reasonable, some not) ▪ Support for bboying music* ▪ an iOS version ▪ Cloud storage - “MyTricks everywhere” ▪ hoyga usted se puede acer millionaire con este projecto todo lo que nesesita es mi ayuda dejeme esplicarle porfabor ◦ After some early versions, App was looking much better and got more stable -
Marketshare ◦ Mostly Android 2 ◦ A few Android 4
devices ◦ 5 or 6 Android 3 tablets ◦ Entry-level devices ◦ no money -
Scaling, next steps ◦ First implementations ◦ Credits ◦ Backup
◦ “Quote of the day” ◦ First social implementation - Calendar ◦ Didn’t want to depend on Internet ◦ Parse traffic limitation ( 1M requests / month! ) ◦ RULE: Altruist project - no money involved -
First mistakes when becoming social ◦ Credits showing “visit FB
site to appear here” ◦ Quote of the day saying “tweet us your quote of the day” ◦ Very low impact ◦ Direct messages to friends gave much more results -
First mistakes when becoming social ◦ Credits showing “visit FB
site to appear here” ◦ Quote of the day saying “tweet us your quote of the day” ◦ Very low impact ◦ Direct messages to friends gave much more results -
First mistakes when becoming social ◦ Solution: Directly posting Messages
on the App ◦ Big effort to eliminate internet dependency ◦ Big effort to save Parse requests ◦ Caching ◦ Access control ◦ Big effort on moderation / avoid trolling ◦ Prepairing the App for high usage rates ◦ “The QR anecdote” -
Results ◦ Very few usage of the social features ◦
Very few social network activity ◦ App is being used as standalone ◦ Downloads increase, but… ◦ Active installs are frozen, or decreasing -
… And so went 2014 ◦ Mass of standalone users
◦ Not knowing about each other (neither do I) ◦ App is not dead, but not as alive as I’d like - “Zombie” -
… And so went 2014 ◦ Mass of standalone users
◦ Not knowing about each other (neither do I) ◦ App is not dead, but not as alive as I’d like - “Zombie” -
First successful steps ◦ Successful social implementations (more interest in
the App) ◦ Chat (Quotes with response) ◦ Bboying music!! (SoundCloud hack) ◦ Bboying Map!! ◦ Push support, accurate segmentation ◦ “News indicators” ◦ Started UI Redesign (strictly necessary?) ◦ Not worrying about traffic anymore -
Results come, be patient... ◦ More installs ◦ Updates come
slowly, but are coming -
Good news ◦ Much more usage of the App! ◦
Good (and bad) feedback ◦ French translation! -
Keys that brought success Stay ~one year behind the latest
(widespread marketshare) Don’t rush with the (big) changes Don’t marry the wrong person (libs) Your reputation costs more than 40$ DONE is better than PERFECT (Sometimes) -
Keys that brought success Stay ~one year behind the latest
(widespread marketshare) Don’t rush with the (big) changes Don’t marry the wrong person (libs) Your reputation costs more than 40$ Sometimes DONE is better than PERFECT -
Stay ~one year behind the latest No Material UI (obviously)
No Wearable support (2012-2013) No ActionBar support (see “don’t marry the wrong person”) -
Keys that brought success Stay ~one year behind the latest
(widespread marketshare) Don’t rush with the (big) changes Don’t marry the wrong person (libs) Your reputation costs more than 40$ Sometimes DONE is better than PERFECT -
Don’t rush with the big changes ORM layer refactoring? DONE
is better than PERFECT Camera pictures support - Duplicated impl. CLEAN, MVP, DI -
Keys that brought success Stay ~one year behind the latest
(widespread marketshare) Don’t rush with the (big) changes Don’t marry the wrong person (libs) Your reputation costs more than 40$ Sometimes DONE is better than PERFECT -
Don’t marry the wrong person Handmade ORM layer (no external
libs) • AndroidRecord • SugarORM • ORMLite • GreenDAO No usage of ActionBarSherlock (past) Soft ActionBarCompat integration dagger rules, but… dagger2 came l8r -
Keys that brought success Stay ~one year behind the latest
(widespread marketshare) Don’t rush with the (big) changes Don’t marry the wrong person (libs) Your reputation costs more than 40$ Sometimes DONE is better than PERFECT -
Your reputation costs more than 40$ Reject agressive commercial campaigns
Don’t try to sell rubbish Don’t bother your users with commercial stuff -
App lifecycle tips I blow ya’ head off hey wassup
with you wat u sayin bout me ?!?! At least try to get a DECENT ZOMBIE! -
Keeping your App safe ◦ Adopt & follow naming conventions
◦ Always depend on abstractions, not concretions ◦ CLEAN your code - Separate responsibilities ◦ Cover everything with Tests -
Keeping your App safe ◦ Adopt & follow naming conventions
◦ Always depend on abstractions, not concretions ◦ CLEAN your code - Separate responsibilities ◦ Cover everything with Tests -
Naming conventions: Example Use case: Listing Customers CustomersActivity CustomersFragment activity_customers.xml
fragment_customers.xml @+id/customers_list row_customer.xml @+id/customer_txt_title @+id/customer_img_thumbnail -
Naming conventions: Important thing They bring execution-time errors to compilation
time!! row_customer.xml @+id/txt_title row_event.xml @+id/txt_title findViewById(R.id.txt_title) row_customer.xml @+id/customer_txt_title row_event.xml @+id/event_txt_title findViewById(R.id.customer_txt_title) Danger. CHECK !! Safer -
Keeping your App safe ◦ Adopt & follow naming conventions
◦ Always depend on abstractions, not concretions ◦ CLEAN your code - Separate responsibilities ◦ Cover everything with Tests -
Keeping your App safe ◦ Adopt & follow naming conventions
◦ Always depend on abstractions, not concretions ◦ CLEAN your code - Separate responsibilities ◦ Cover everything with Tests -
CLEAN your code - separate resp. ◦ MVP, MVVM ◦
CustomViews ◦ Interactor Pattern ◦ DataSources ◦ Retrofit ◦ Gson, SimpleXML ◦ Pass dependencies on constructor ◦ Code smells ◦ RecyclerView ◦ Boundaries ◦ etc. @ voghDev -
Keeping your App safe ◦ Adopt & follow naming conventions
◦ Always depend on abstractions, not concretions ◦ CLEAN your code - Separate responsibilities ◦ Cover everything with Tests … A very desirable TODO -
TagCloud: Android miscellaneous Clean code SOLID Builder Pattern parcelabler WTF
/ min RxJava Material Design Renderers Lazy Parsing Model-View-Presenter DataBinding RecyclerView Dalton Pattern ViewModel Mocking Live Templates Boilerplate code ButterKnife Z-Vectoring Kitt Calls Bitmap recycling Mortar Flow PorterDuff GeoPoint Reflection CardView -
TagCloud: Android miscellaneous Clean code SOLID Builder Pattern parcelabler WTF
/ min RxJava Material Design Renderers Lazy Parsing Model-View-Presenter DataBinding RecyclerView Dalton Pattern ViewModel Mocking Live Templates Boilerplate code ButterKnife Z-Vectoring Kitt Calls Bitmap recycling Mortar Flow PorterDuff GeoPoint Reflection CardView WTF ? ? wtf is dat sheet -
And that’s all folks Q & A & T Questions
& Answers & Trolling -
And that’s all folks Q & A & T Questions
& Answers & Trolling Una vaina looocaaaa.... -
And that’s all folks Q & A & T Questions
& Answers & Trolling