QuineDB is a quine that is also a key/value store.
If your database can't print its own source code, can you really trust it?
QuineDB consists of the quinedb
script in this repository. It
is written in Bash and requires Bash 4.
When you run it, the (possibly modified) source code of quinedb
is
printed to STDOUT
, and the results of the specific command run are
printed to STDERR
. Therefore, each time you run a write operation
you must redirect STDOUT
to an appropriate place. For consistency,
we recommend doing this for all operations. However this can be very
tedious to do directly, so a bash function such as the following can
be helpful, and the examples that follow will make use of it:
qdb () {
./quinedb "$@" > qdb.out 2> qdb.err || return "$?"
cat qdb.out > quinedb
cat qdb.err
rm qdb.err qdb.out
}
QuineDB has four commands: get
, set
, delete
, and keys
.
To set a key to a value, use quinedb set <k> <v>
:
$ qdb set foo bar
OK
$ qdb set count 42
OK
To get a value, use quinedb get <k>
:
$ qdb get foo
bar
To list the keys in the database, use quinedb keys
:
$ qdb keys
count
foo
To delete a key, use quinedb delete <k>
:
$ qdb delete foo
OK
$ qdb keys
count
Keys and values are printed in the syntax of bash strings, which
allows keys
to print one key on each line unambiguously (even if a
key contains a newline) and get
to unambiguously print no output for
a missing key.
$ qdb set \
$'this\nkey\nhas\nfour\nlines' \
$'and the \'value\' has \"quotes\"'
OK
$ qdb set empty ''
OK
$ qdb keys
$'this\nkey\nhas\nfour\nlines'
empty
$ qdb get $'this\nkey\nhas\nfour\nlines'
$'and the \'value\' has \"quotes\"'
$ qdb get empty
$''
$ qdb get missing
To group several operations into an atomic transaction you can simply
chain operations by redirecting STDOUT
to an invocation of
/usr/bin/env bash
like so:
$ ./quinedb set k1 v1 | \
/usr/bin/env bash -s set k2 v2 | \
/usr/bin/env bash -s set k3 v3 | \
/usr/bin/env bash -s keys > tmp; chmod +x tmp; mv tmp quinedb
OK
OK
OK
k1
k3
k2
- Why shouldn't your database be a quine?
- If your data and the database code are not stored in the same place, you risk losing track of one, making the other useless.
- You can store various versions of your database, or even fork it, with no extra effort.
I was able to insert 100 k/v pairs into an empty database in a mere 10 seconds. The runtime of each operation is (probably) proportional to O(n·log(n)), so it's not too surprising that inserting 1000 k/v pairs took over 11 minutes. If you need the database to be fast, we recommend not putting too much data in it.
I'm not sure. How does bash work exactly?
I'm not sure you've been paying attention.
Well I mean, um.
Pull requests are welcome.
- @timgaleckas for reviewing and coming up with the mechanism for transactions
Copyright © 2016 Gary Fredericks
Distributed under the Eclipse Public License either version 1.0 or (at your option) any later version.