refORM: Death to ORMs in .NET
Download as PPTX, PDFâ˘5 likesâ˘4,926 views
This document discusses the disadvantages of object-relational mappers (ORMs) and promotes using micro ORMs instead. It argues that ORMs introduce unnecessary abstraction and complexity, forcing developers to work around the ORM rather than having the ORM work for them. Micro ORMs like Dapper, ORMLite, Massive, PetaPoco, and Simple.Data are presented as alternatives that are closer to the metal and impose less overhead than traditional heavyweight ORMs. Code examples are provided to demonstrate basic data access tasks with each micro ORM.
refORM: Death to ORMs in .NET
- 1. R E F O R MDeath to ORMs in .NET
- 2. âA Propos de Moiâ@kouphaxhttp://github.com/kouphaxhttp://yobriefca.se
- 3. DisclaimerThis presentation is full of opinion. My opinion. The fact is ORMs have been around for a long time and will be for a while yet. If things were as simple as I portray them in this presentation you wouldnât be using them today. But you probably are.All I ask is that you never blindly use an ORM or any technology without prior brain thinkingâ˘
- 4. On with the showâŚ
- 5. ORMs are for LifeâŚSo youâve adopted an ORM, possiblyNHibernateMyBatisEntity FrameworkThe samples, blogs and demos promise you a better life, your âhello worldâ grade apps are easy and it adds another shiny buzzword to your CV (Resume Driven Development)You dive right in investing a heap of development cycles, things start out wellâŚ
- 6. Suddenly and Without WarningâŚ
- 7. The Cracks Start To ShowLeaky AbstractionsBugger to debugForced to work for the ORM rather than the ORM working for youDBAs begin to call you namesCode generation and heavy use of model-polluting attributes bog you downCost of change is highOverreliance on ORMs make you dumb
- 9. ORMs in .NET â More Like WTFsNHibernateHQL â WTF?MyBatisXML as a medium for SQL? â WTF?Entity FrameworkGreat demoware but the reality is more like WTF?
- 10. Lets Be HonestHow many features of a heavyweight ORM do you really use regularly? 80%? 20%? 10%?How many of these features exist purely to support the abstraction?How difficult is the mapping between DBâs and objects really? Isnât SQL the best DSL for working with data in a relational database?
- 11. So why is this abstraction necessary?
- 12. It isnât.
- 13. Historically speaking the alternatives were pretty slim. Most people would need to roll their own project specific data access layerMany ORMs grew out of that sort of thinking but from a community that thought XML was sexySome people are weirdThe world is a saner place these days. Reflection is fasterStatically typed languages have embraced dynamics, and,most people realise XML is a noisy medium for almost everything
- 14. The Advent of Micro ORMs
- 15. Level of AbstractionMicro ORMsHeavyweight ORMs
- 16. Dapper
- 17. DapperCreated by Sam Saffron of StackoverflowSingle file (available on Nuget)Set of extension methods to IDbConnectionSupports most .NET ADO providersSQL Server, Oracle, SQLite, MySQL, SQL CE Works on boths POCOs and dynamic objectsPretty close to the metalNo connection handlingNo helpers for INSERT, DELETE, UPDATE etc.Community extensions for CRUD operations.NET 3.5 support
- 18. Demo
- 19. QueriesDapper
- 20. Inserts, Updates & DeletesDapper
- 21. Inserts, Updates & Deletes (CE)Dapper
- 22. Dapper is very close to the metalVery little in the way of providing helpers for 1..N, N..N relationshipsDapper Community Extensions bridge some of the mental gap for people not used to SQLDapperAnd the restâŚ
- 23. ORMLite
- 24. ORMLitePart of the ServiceStack technology suiteSet of CRUD extensions around System.Data.* namespaceCurrently supports SQLite and SQL ServerMaps data to and from concrete POCOsVia conventionVia attributes on the POCO â preferredVia raw SQLRelies heavily on POCO attributionSupports .NET 3.5
- 25. Demo
- 27. QueriesORMLite
- 28. Inserts, Updates & DeletesORMLite
- 29. ORMLite also allows you to create tables from entitiesEach model must have an Id property which must be its primary key.Integrates into the rest of the ServiceStack technology stackORMLiteAnd the restâŚ
- 30. Massive
- 31. MassiveCreated by Rob ConreySingle file (available on Nuget)Makes extensive use of dynamicsNo mapping to POCOs for queriesPrefers to wrap tables rather than entitiesProvides a single point of configurationNo need for attributesOOTB Paging support.NET 4 Only
- 32. Demo
- 33. Creating the Dynamic ModelMassive
- 34. QueriesMassive
- 35. Inserts, Updates & DeletesMassive
- 36. ValidationsEventsAlso Github saysâŚAsynchronous support for queriesAbility to reflect Schema MetadataAbility to generate a (DB) default rowFactory method for ad hoc/on the fly queriesMassiveAnd the restâŚ
- 37. PetaPoco
- 38. PetaPocoYet another single file (and on Nuget)Inspired by Massive and SubsonicSupports POCOs (and optional attributes)Paging and transaction support OOTBWorks with SQL Server, SQL Server CE, MySQL, PostgreSQL and Oracle..NET 3.5 upwards
- 39. Demo
- 41. QueriesPetaPoco
- 42. Inserts, Updates & DeletesPetaPoco
- 43. Out of the box support for 1..N, N..N fetching of objectsSlightly awkward approachRelationExtensions project abstracts this awkwardness for 1..N, N..1 queriesProvides a set of T4 Templates for generating POCOs from DB schemaPetaPocoAnd the restâŚ
- 44. Simple.Data
- 45. Simple.DataSlightly different beast though itâs principles are firmly in the MicroORM camp.Data Access ComponentNot just for relational databasesNoSQL supportExtensible Adapter/Provider supportSupports a wide range of data storesSQL Server, SQL Server CE, Oracle, MySQL, SQLite, MongoDB.NET 4 onlyExtensive use of dynamics and meta-programming concepts.TryInvokeMember, Method name reflection
- 46. Demo
- 49. Inserts, Updates & DeletesSimple.Data
- 50. Inclusion of Simple.Data here may seem like a bit of contradiction. After all it is a straight up abstraction.It makes the 95th centile of all CRUD tasks âsimpleâAbility to compose DAL from other lightweight frameworks for the other edge case tasksAdditive approach rather than upfront monolithic approachSimple.DataAnd the restâŚ
- 51. QUESTIONS?
Editor's Notes
- #2: The purpose of this presentation today is to try and show you that the typical ORMs and Data Mappers you probably use in your projects are not the only option. Iâm going to touch on some of their common pain points and hopefully most of you will have experienced some of them in some shape or form. After that Iâll give you a fairly high level overview of many of the more lightweight appraoches which Iâm hoping will speak for themselves in terms of ease of use and brevity.
- #5: The Answer
- #6: What are they?What exists out there?
- #7: The Answer
- #8: With ORMs you get certain things for free with certain sacrifices. Many of these sacrifices directly affect your DBAsYouâve immediatley created a defensive DBA and that aint pretty
- #9: Eventually you realise your ORM has let itself go and no amount of refactoring is going to change things
- #10: Iâve had to
- #12: The question
- #13: The Answer
- #15: The modern development landscape has created a melting pot of lightweight innovation and micro ORMs are becoming quite prominent. The name itself is often a bit of misnomer. I suppose they are closer to Data Mappers than true ORMs but itâs a catchy name so
- #17: The modern development landscape has created a melting pot of lightweight innovation and micro ORMs are becoming quite prominent. The name itself is often a bit of misnomer. I suppose they are closer to Data Mappers than true ORMs but itâs a catchy name so
- #19: The modern development landscape has created a melting pot of lightweight innovation and micro ORMs are becoming quite prominent. The name itself is often a bit of misnomer. I suppose they are closer to Data Mappers than true ORMs but itâs a catchy name so
- #24: The modern development landscape has created a melting pot of lightweight innovation and micro ORMs are becoming quite prominent. The name itself is often a bit of misnomer. I suppose they are closer to Data Mappers than true ORMs but itâs a catchy name so
- #25: Attributes are fine(ish)
- #26: The modern development landscape has created a melting pot of lightweight innovation and micro ORMs are becoming quite prominent. The name itself is often a bit of misnomer. I suppose they are closer to Data Mappers than true ORMs but itâs a catchy name so
- #31: The modern development landscape has created a melting pot of lightweight innovation and micro ORMs are becoming quite prominent. The name itself is often a bit of misnomer. I suppose they are closer to Data Mappers than true ORMs but itâs a catchy name so
- #33: The modern development landscape has created a melting pot of lightweight innovation and micro ORMs are becoming quite prominent. The name itself is often a bit of misnomer. I suppose they are closer to Data Mappers than true ORMs but itâs a catchy name so
- #34: You can also do things like set the table name
- #35: You can also do things like set the table name
- #38: The modern development landscape has created a melting pot of lightweight innovation and micro ORMs are becoming quite prominent. The name itself is often a bit of misnomer. I suppose they are closer to Data Mappers than true ORMs but itâs a catchy name so
- #40: The modern development landscape has created a melting pot of lightweight innovation and micro ORMs are becoming quite prominent. The name itself is often a bit of misnomer. I suppose they are closer to Data Mappers than true ORMs but itâs a catchy name so
- #41: You can also do things like set the table name
- #42: You can also do things like set the table name
- #45: The modern development landscape has created a melting pot of lightweight innovation and micro ORMs are becoming quite prominent. The name itself is often a bit of misnomer. I suppose they are closer to Data Mappers than true ORMs but itâs a catchy name so
- #47: The modern development landscape has created a melting pot of lightweight innovation and micro ORMs are becoming quite prominent. The name itself is often a bit of misnomer. I suppose they are closer to Data Mappers than true ORMs but itâs a catchy name so
- #48: You can also do things like set the table name
- #49: You can also do things like set the table name
- #51: But a last it isnât a swathe of mandatory configuration. Convention based.But itâs still very lightweight in comparison to the likes And the philosophy rests firmly in the MicroORM worldMicroORMs are more composable due to their sze and complexity