Comments:
http://codereview.chromium.org/131100/diff/1/3 File src/compilation-cache.cc (right): http://codereview.chromium.org/131100/diff/1/3#newcode58 Line 58: The rest of the file has two newlines between the definitions. http://codereview.chromium.org/131100/diff/1/3#newcode68 Line 68: static Object* tables[NUMBER_OF_TABLE_ENTRIES] = { 0, }; Maybe this should really be an array of pointers to an array of generations? I think we should refactor the code so we have a CompilationCacheEntry per type that has an array of generations (in the form of CompilationCacheTables). http://codereview.chromium.org/131100/diff/1/3#newcode115 Line 115: if (result->IsJSFunction()) { No promotion of hits in older generation to newer generation? http://codereview.chromium.org/131100/diff/1/3#newcode142 Line 142: if (result->IsFixedArray()) { No promotion of hits in older generation to new generations? http://codereview.chromium.org/131100/diff/1/3#newcode189 Line 189: int generation = 0; Now that the generation count isn't included in the Entry, you should probably just get rid of the comment. http://codereview.chromium.org/131100/diff/1/3#newcode195 Line 195: while (generation < generations[SCRIPT]) { It feels like we should be able to somehow share more of the generation-based lookup code between the different entry kinds. There's a lot of repeated code (not all your fault though). http://codereview.chromium.org/131100 --~--~---------~--~----~------------~-------~--~----~ v8-dev mailing list [email protected] http://groups.google.com/group/v8-dev -~----------~----~----~----~------~----~------~--~---
