A package (hashgen) that supplies a crypto hash server and utilties for reuse.
Linux supported. Depends on:
- Go standard library
- OS commands
uuidgenandgrep
See project jchash for usage.
The hashgen package has tests.
$ cd hashgen
$ go test -v
Sample coverage check:
$ cd hashgen
$ go test -coverprofile=cover.out
...
$ go tool cover -func=cover.out
github.com/harapr-jc/hashgen/dao.go:24: New 100.0%
github.com/harapr-jc/hashgen/dao.go:29: Append 71.4%
github.com/harapr-jc/hashgen/dao.go:65: Get 100.0%
github.com/harapr-jc/hashgen/hasher.go:15: getSalt 80.0%
github.com/harapr-jc/hashgen/hasher.go:26: getCryptoHash 100.0%
github.com/harapr-jc/hashgen/lru.go:43: NewCache 100.0%
github.com/harapr-jc/hashgen/lru.go:59: Add 84.6%
github.com/harapr-jc/hashgen/lru.go:84: Evict 88.9%
github.com/harapr-jc/hashgen/lru.go:98: Get 94.1%
github.com/harapr-jc/hashgen/server.go:51: HandleGetHashRequest 100.0%
github.com/harapr-jc/hashgen/server.go:73: HandleHashRequest 93.1%
github.com/harapr-jc/hashgen/server.go:133: HandleStatsRequest 100.0%
github.com/harapr-jc/hashgen/server.go:138: StartServer 90.3%
github.com/harapr-jc/hashgen/stats.go:38: Accumulate 72.7%
github.com/harapr-jc/hashgen/stats.go:59: GetJson 85.7%
github.com/harapr-jc/hashgen/uuid.go:11: getUuid 75.0%
total: (statements) 88.4%
Some of the tests support benchmarks.
$ cd hashgen
$ go test -bench=".*"
The use of uuid instead of monotonically increasing id for a job identifier allows server restart without having to persist the next id. Furthermore, no synchronization is required on next id.
A drawback is that, without a database back end, ordering of uuid is less efficient for persistent storage. A consequence is that records that are evicted from the cache will take longer to find.
The purpose for using the cache is to put an upper bound on memory usage for the server.
The design intent is that n most recently used jobs are held in cache, all are passed thru to persistent storage. On a cache miss, job data could be fetched from persistent storage. The default backing file is called 'backup.json'. One would prefer to use a performant data store instead. The Go standard library has sql API, but no database drivers are included.
- Raise code coverage
- Consider API version scheme
- Consider implementing password policies
- Add safety for when backing file size approaches partition size