What do you mean, Backwards Compatibility?

Transcription

1 #gotober What do you mean, Backwards Compatibility? Trisha Gee, Java Driver Friday, 18 October 13

2 What s the problem?

3 The Domain

4 MongoDB is an open-source document database, featuring: Document-Oriented Storage Full Index Support Replication & High Availability Auto-Sharding Querying Fast In-Place Updates Map/Reduce GridFS

5 The Driver

6

7

8

9 Why is this my problem? Friday, 18 October 13

10 What do we want to do?

11 Design Goals Friday, 18 October 13 Intuitive API Consistency Understandable exceptions Cleaner design Test friendly Backwards compatible

12 Happy Users

13 Three Types Of Users

14 Three Types Of Users 1. Java Developers Friday, 18 October 13

15 Three Types Of Users 1. Java Developers 2. ODMs / other drivers / third parties Friday, 18 October 13

16 Three Types Of Users 1. Java Developers 2. ODMs / other drivers / third parties 3. Contributors Friday, 18 October 13

17 What s stopping us?

18 Lots of unknowns

19 Lots of APIs

20 Backward Compatibility

21 What did we do?

22 Architecture

23 How Hard Can It Be?

24 Domain

25 MongoDB is an open-source document database, featuring: Document-Oriented Storage Full Index Support Replication & High Availability Auto-Sharding Querying Fast In-Place Updates Map/Reduce GridFS

26 Bit more complex...

27 MongoDB is an open-source document database, featuring: Document-Oriented Storage Full Index Support Replication & High Availability Auto-Sharding Querying Fast In-Place Updates Map/Reduce GridFS

28 Instead of this...

29 Model the Domain

30 Scala Driver

31 Backwards Compatible!

32 MongoDB is an open-source document database, featuring: Document-Oriented Storage Full Index Support Replication & High Availability Auto-Sharding Querying Fast In-Place Updates Map/Reduce GridFS

33 Multi Server

34 Replica Set

35 Replica Set

36

37 Replica Sharding Set Friday, 18 October 13

38

39 Finding Servers the Old Way Friday, 18 October 13

40

41

42

43

44

45

46

47

48

49

50 But now...

51

52

53

54

55 MongoDB is an open-source document database, featuring: Document-Oriented Storage Full Index Support Replication & High Availability Auto-Sharding Querying Fast In-Place Updates Map/Reduce GridFS

56 MongoDB is an open-source document database, featuring: Document-Oriented Storage Full Index Support Replication & High Availability Auto-Sharding Querying Fast In-Place Updates Map/Reduce GridFS

57 The New API

58 Caveats It won t look like this Haven t decided consistent names yet Need something that suits all drivers Friday, 18 October 13

59 MongoDB is an open-source document database, featuring: Document-Oriented Storage Full Index Support Replication & High Availability Auto-Sharding Querying Fast In-Place Updates Map/Reduce GridFS

60 patron = { _id: "joe", name: "Joe Bookreader", address: { street: "123 Fake St", city: "Faketon", state: "MA", zip: } books: [ 27464, ,...] }

61 List books = new BasicDBList(); books.add(27464, ); new BasicDBObject("_id", "joe").append("name", "Joe Bookreader").append("address", new BasicDBObject("street", "123 Fake St").append("city", "Faketon").append("state", "MA").append("zip", 12345)).append("books", books); collection.insert(books); Building a Document Friday, 18 October 13

62 DBCollection collection = database.getcollection("coll"); ArrayList<Patron> resultstoreturn = new ArrayList<Patron>(); DBObject query = new BasicDBObject("name", thenametofind); DBCursor results = collection.find(query); for (DBObject dbobject : results) { Patron patron = new Patron((String) dbobject.get("name"), new Address((String)dbObject.get("street"), (String)dbObject.get("city"), (String)dbObject.get("state"), (Integer)dbObject.get("zip")), (BasicDBList)dbObject.get("books")); resultstoreturn.add(patron); } return resultstoreturn; Getting it back Friday, 18 October 13

63 DBCollection collection = database.getcollection("coll"); ArrayList<Patron> resultstoreturn = new ArrayList<Patron>(); DBObject query = new BasicDBObject("name", thenametofind); DBCursor results = collection.find(query); for (DBObject dbobject : results) { Patron patron = new Patron((String) dbobject.get("name"), new Address((String)dbObject.get("street"), (String)dbObject.get("city"), (String)dbObject.get("state"), (Integer)dbObject.get("zip")), (BasicDBList)dbObject.get("books")); resultstoreturn.add(patron); } return resultstoreturn; Casting is fun Friday, 18 October 13

64 MongoCollection<Patron> collection = database.getcollection("coll", new PatronCodec()); Document query = new Document("name", thenametofind); return collection.find(query).into(new ArrayList<Patron>()); New API Friday, 18 October 13

65 MongoCollection<Patron> collection = database.getcollection("coll", new PatronCodec()); Document query = new Document("name", thenametofind); return collection.find(query).into(new ArrayList<Patron>()); Separation of concerns Friday, 18 October 13

66 MongoDB is an open-source document database, featuring: Document-Oriented Storage Full Index Support Replication & High Availability Auto-Sharding Querying Fast In-Place Updates Map/Reduce GridFS

67 MongoDB is an open-source document database, featuring: Document-Oriented Storage Full Index Support Replication & High Availability Auto-Sharding Querying Fast In-Place Updates Map/Reduce GridFS

68 Find

69 collection.find(query).skip(1000).limit(100); Find Friday, 18 October 13

70 collection.find(query).skip(1000).limit(100); collection.find(query).skip(1000).limit(100); Find Friday, 18 October 13

71 collection.find(query).skip(1000).limit(100); collection.find(query).skip(1000).limit(100); collection.find(query, fields); Find Friday, 18 October 13

72 Which One?

73 Find

74 collection.find(query).skip(1000).limit(100); collection.find(query).skip(1000).limit(100); collection.find(query, fields); Find Friday, 18 October 13

75 collection.find(query).skip(1000).limit(100); collection.find(query).skip(1000).limit(100); collection.find(query, fields); collection.find(query).project(fields); Find Friday, 18 October 13

76 Fewer Decisions

77 Ctrl + space friendly

78 Remove

79 collection.remove(query); Remove Friday, 18 October 13

80 collection.remove(query); collection.find(query).remove(); Remove Friday, 18 October 13

81 Update

82 collection.update(query, newvalues) Update Friday, 18 October 13

83 collection.update(query, newvalues) collection.find(query).updateone(newvalues); Update Friday, 18 October 13

84 Overloaded Methods

85 Update

86 collection.update(query, newvalues); collection.find(query).updateone(newvalues); collection.update(query, newvalues, false, false, JOURNALED); Update Friday, 18 October 13

87 collection.update(query, newvalues); collection.find(query).updateone(newvalues); collection.update(query, newvalues, false, false, JOURNALED); collection.find(query).withwriteconcern(journaled).updateone(newvalues); Update Friday, 18 October 13

88 Update

89 collection.update(query, newvalues); collection.find(query).updateone(newvalues); collection.update(query, newvalues, false, false, JOURNALED); collection.find(query).withwriteconcern(journaled).updateone(newvalues); collection.update(query, newvalues, true, false, JOURNALED); Update Friday, 18 October 13

90 collection.update(query, newvalues); collection.find(query).updateone(newvalues); collection.update(query, newvalues, false, false, JOURNALED); collection.find(query).withwriteconcern(journaled).updateone(newvalues); collection.update(query, newvalues, true, false, JOURNALED); collection.find(query).withwriteconcern(journaled).upsert().updateone(newvalues); Update Friday, 18 October 13

91 Update

92 collection.update(query, newvalues); collection.find(query).updateone(newvalues); collection.update(query, newvalues, false, false, JOURNALED); collection.find(query).withwriteconcern(journaled).updateone(newvalues); collection.update(query, newvalues, true, true, JOURNALED); collection.find(query).withwriteconcern(journaled).upsert().update(newvalues); Update Friday, 18 October 13

93 MongoDB is an open-source document database, featuring: Document-Oriented Storage Full Index Support Replication & High Availability Auto-Sharding Querying Fast In-Place Updates Map/Reduce GridFS

94 MongoDB is an open-source document database, featuring: Document-Oriented Storage Full Index Support Replication & High Availability Auto-Sharding Querying Fast In-Place Updates Map/Reduce GridFS

95 Atomic Operations

96 Find and Modify

97 collection.findandmodify(query, newvalues); Find and Modify Friday, 18 October 13

98 collection.findandmodify(query, newvalues); collection.find(query).getoneandupdate(newvalues); Find and Modify Friday, 18 October 13

99 They hate me!

100 Find and Modify

101 collection.findandmodify(query, update); collection.find(query).getoneandupdate(update); collection.findandmodify(query, fields, criteria, false, newvalues, false, false); Find and Modify Friday, 18 October 13

102 collection.findandmodify(query, update); collection.find(query).getoneandupdate(update); collection.findandmodify(query, fields, criteria, false, newvalues, false, false); collection.find(query).project(fields).sort(criteria).getoneandupdate(newvalues); Find and Modify Friday, 18 October 13

103 collection.findandmodify(query, update); collection.find(query).getoneandupdate(update); collection.findandmodify(query, fields, criteria, false, newvalues, true, false); collection.find(query).project(fields).sort(criteria).updateoneandget(newvalues); Find and Modify Friday, 18 October 13

104 Consistency

105 collection.find(query).count(); collection.find(query).remove(); collection.find(query).update(newvalues); collection.find(query).updateoneandget(newvalues); collection.find(query).getoneandupdate(newvalues); collection.find(query).sort(sortcriteria).skip(9).limit(10).get(); collection.find(query).sort(sortcriteria).skip(9).limit(10).getone(); collection.find(query).sort(ascending("name")).getone(); Consistency at last Friday, 18 October 13

106 How do we know we did it?

107 Tests Pass...

108 ...and more tests pass... Friday, 18 October 13

109 ...even more tests pass... Friday, 18 October 13

110 It s being used in anger Friday, 18 October 13

111 Conclusion

112 Conclusion

113 Conclusion Model your domain Friday, 18 October 13

114 Conclusion Model your domain Know your users Friday, 18 October 13

115 Conclusion Model your domain Know your users API design is hard Friday, 18 October 13

116 Questions Friday, 18 October 13

117 MongoFuture<Document> future = collection.find(query).sort(criteria).skip(9).limit(10).asyncone(); MongoFuture<Long> count = collection.find(query).asynccount(); MongoFuture<WriteResult> replaceresult = collection.find(query).asyncreplace(replacement); What about async? Friday, 18 October 13

118 Understandable Exceptions Client Exceptions Server Exceptions No more parsing error Strings Friday, 18 October 13

119 Documentation Self documenting code JavaDoc Unit, Functional and Acceptance Tests Blogs Tutorials Friday, 18 October 13

120 My Questions

121 1. Are you using the Java driver? Friday, 18 October 13

122 2. What do you like about it? Friday, 18 October 13

123 3. What are your pain points? Friday, 18 October 13