| Anonymous | Login | Signup for a new account | 09-02-2010 20:15 UTC |
| Main | My View | View Issues | Change Log | Docs |
| Viewing Issue Simple Details [ Jump to Notes ] | [ View Advanced ] [ Issue History ] [ Print ] | |||||||||||
| ID | Category | Severity | Reproducibility | Date Submitted | Last Update | |||||||
| 0001876 | [Squeak] Collections | minor | always | 10-06-05 18:45 | 10-03-09 19:34 | |||||||
| Reporter | MarcusDenker | View Status | public | |||||||||
| Assigned To | andreas | |||||||||||
| Priority | normal | Resolution | fixed | |||||||||
| Status | assigned | Product Version | 3.9 | |||||||||
| Summary | 0001876: [ENH] identityHash | |||||||||||
| Description |
In 2001, Scott A Crosby posted a couple of changesets to fix the bad scalability of collections that use identityhash. The idea: put the primitive into a method "hashBits", then define "identityHash" as hashBits*primNumber. So we do the upscaling directly in the hash method, not later in scanFor: The good effect: scanFor: does not need to test for size, all object that use identityHash get useable hash values. Negative: #identityHash now is slower due to the multiplication... but that should not be noticable. Phillipe did a test of the idea by implementing #hash like that and using OrderdCollection. The picture attached show the result. TODO: check if such a big scale is needed... maybe a smaller number is enough? |
|||||||||||
| Additional Information | ||||||||||||
| Attached Files |
|
|||||||||||
|
|
||||||||||||
| Mantis 1.0.8[^]
Copyright © 2000 - 2007 Mantis Group
88 total queries executed. 42 unique queries executed. |