ThePerformance Evaluation ofControl
John P. Kearns · 1982
A methodolo implementation ofcontro methodology specifies ad driving simulations of4 metric forperformance ei separation ofactivity whi fromthat whichisnot. Themethodology isap schemesforcoroutines. context ofthegeneralize manceimprovements ove agedynamically. A new basis ofthese results. IndexTenns-Coroutii spaghetti stack, storage m: gyfortheexperimental investigation oftheistypically usedinarecursive setting. Asintermodule control Ainprogramming languages ispresented. The flowinalogically parallel regime isnotrestricted tobeLIFO, Lata-collection technique, theenvironment forthestorage associated with a'program module hastobere- control implementations, anda meaningful valuation. Thekeyinthemethodology isthetamed whencontrol exits themodule. ich isaffected bythecontrol implementation A numberofstorage management schemes havebeenpro- posed andusedtoimplement retentive control. These include oplied inthecontext oftemporary allocation various versions ofheaps, spaghetti stacks (51, andacombina- Preallocation oftemporary storage inthe dstack isshown tolead tosignificant perfor- tionofstructures (3),(4).Eachhascertain advantages and brallocating anddeallocating temporary stor-disadvantages overthe' others anditisunclear whichone coroutine implementation issuggested ontheshould beusedgiven particular control formsanddisciplines. Modifying ascheme totailor ittoacontrol environment could nes, measurement, performance evaluation, perhaps improve its performance. anagement. Duetothis difficulty ofknowing whentochoose onescheme overanother intheimplementation ofalanguage containing retentive control, this paper develops amethodology that can