Documented by mathew email@example.com.
The exact library used depends on how Ruby was compiled. It could be any of the following:
The original ndbm library is released in 4.3BSD. It is based on dbm library in Unix Version 7 but has different API to support multiple databases in a process.
Berkeley DB versions 1 thru 5, also known as BDB and Sleepycat DB, now owned by Oracle Corporation.
Berkeley DB 1.x, still found in FreeBSD and OpenBSD.
gdbm the GNU implementation of dbm.
All of these dbm implementations have their own Ruby interfaces available, which provide richer (but varying) APIs.
Before you decide to use DBM, there are some issues you should consider:
Each implementation of dbm has its own file format. Generally, dbm libraries will not read each other’s files. This makes dbm files a bad choice for data exchange.
Even running the same OS and the same dbm implementation, the database file format may depend on the CPU architecture. For example, files may not be portable between PowerPC and 386, or between 32 and 64 bit Linux.
Different versions of Berkeley DB use different file formats. A change to the OS may therefore break DBM access to existing files.
Data size limits vary between implementations. Original Berkeley DB was limited to 2GB of data. Dbm libraries also sometimes limit the total size of a key/value pair, and the total size of all the keys that hash to the same value. These limits can be as little as 512 bytes. That said, gdbm and recent versions of Berkeley DB do away with these limits.
READER = INT2FIX(O_RDONLY|RUBY_DBM_RW_BIT)
WRITER = INT2FIX(O_RDWR|RUBY_DBM_RW_BIT)
WRCREAT = INT2FIX(O_RDWR|O_CREAT|RUBY_DBM_RW_BIT)
NEWDB = INT2FIX(O_RDWR|O_CREAT|O_TRUNC|RUBY_DBM_RW_BIT)
VERSION = rb_str_new2(db_version(NULL, NULL, NULL))