| oilio | so what the point of hashes, can’t we just use void pointers to key/value ? |
| rindolf | oilio: hashes as in hash tables? |
| oilio | yes |
| rindolf | oilio: OK. |
| rindolf | oilio: they are one way to efficiently implement the dictionary Abstract Data Type (ADT). |
| rindolf | oilio: a hash table can store more than one key / value pair. |
| oilio | yeah, I read about it in the wikipedia |
| rindolf | oilio: and you can lookup a value based on a key efficiently. |
| arubin | Linear search is good enough for everyone. |
| oilio | rindolf: what if the whole table won’t fit in the memory? |
| oilio | RAM |
| arubin | Swap. |
| arubin | Amazon S3. |
| imlearningyacc | well if no ram and no swap no allocation |
| arubin | We have the whole Internet for our tables. |
| imlearningyacc | well |
| diminoten | ask reddit how that worked out |
| arubin | And remember, /dev/null is web-scale. |
| imlearningyacc | yea, I store all my data in /dev/null |
| rindolf | oilio: then you’ll need to use a more sophisticated (and slower) data structure that can offload to disk. |
| arubin | It is really fast. |
| arubin | I use the Boost libraries for /dev/null too. |
| arubin | And I use async writes to /dev/null. |
| rindolf | arubin: heh. |
| diminoten | don’t want to get into resource contention when using /dev/null |
| * rindolf | uses /dev/null for backups. |
| diminoten | only so much null to go around |