Supervising Offline Partial Evaluation of Logic Programs Using Online Techniques

Abstract

Abstract. A major impediment for more widespread use of offline partial evaluation is the difficulty of obtaining and maintaining annotations for larger, realistic programs. Existing automatic binding-time analyses still only have limited applicability and annotations often have to be created or improved and maintained by hand, leading to errors. We present a technique to help overcome this problem by using online control techniques which supervise the specialisation process in order to detect such errors. We discuss an implementation in the logen system and show on a series of examples that this approach is effective: very few false alarms were raised while infinite loops were detected quickly. We also present the integration of this technique into a web interface, which highlights problematic annotations directly in the source code. A method to automatically fix incorrect annotations is presented, allowing the approach to be also used as a pragmatic binding time analysis. Finally we show how our method can be used for efficiently locating errors with built-ins inside Prolog source code.

Fast offline partial evaluation of logic programs

“…24 Furthermore the generated annotation file is erroneous and could not be used for specialisation. With our new BTA a correct annotation is generated in less than half a second; the ensuing specialisation by logen took 2.4 s. The generated code is very similar to the one obtained using a manually constructed annotation in Section 3 of [26] or in [35]. In fact, it is slightly more precise and with a simple hint (see Sect.…”

“…In fact, it is slightly more precise and with a simple hint (see Sect. 7.3), we were able to reduce specialisation so as to obtain the exact same code as [35] for the main interpreter loop. With ecce it took over 7 minutes to construct a (very large) specialised program.…”

“…There is obviously the potential for a user to override the BTA in such a way that the specialisation process will no longer terminate. Note, however, that one can still use the watchdog mode [35] to pinpoint such errors. Figure 11 contains a selection of benchmarks from Fig.…”

Modular Decompilation of Low-Level Code by Partial Evaluation

“…The focus has been on demonstrating its effectiveness, i.e., that the socalled interpretation layer can be removed from the compiled programs. To achieve effectiveness, offline [17], online [4,13,24] and hybrid [18] PE techniques have been assessed and novel control strategies have been proposed and proved effective [12,3]. Our work starts off from the premise that interpretive decompilation is feasible and effective as proved by previous work and studies further issues which have not been explored before.…”

Decompilation of Java bytecode to Prolog by partial evaluation

“…The focus has been on demonstrating its effectiveness, i.e., that the so-called interpretation layer can be removed from the compiled programs. To achieve effectiveness, offline [29], online [5,20,39] and hybrid [30] PE techniques have been assessed and novel control strategies have been proposed and proven effective [18,4]. Our work starts off from the premise that interpretive decompilation is feasible and effective as proved by previous work and studies further issues which have not been explored yet.…”