帮忙翻译一段英文论文,100分高分急求!!!
帮忙翻译一段英文论文,100分高分急求!!!
Sharp [22] has presented an approach for pre-computing
summaries for type and dependence analysis in library func-
tions which change rarely. Our approach is orthogonal as
it focuses on identifying changes and propagating their ef-
fects across the code under analysis. Rountev et al. [19]
discuss how to compute more concise summaries of libraries
for generic IDE problems. Integrating these results into our
incremental solver is subject to future work. We envision
using summaries when calls into libraries are changed, but
not the libraries themselves, so as not to repropagate the
changes through the complete library.
Tripp et al. [25] ex0eciently update data-
ow results by
replacing the whole-program control-
ow graph and pointer
analysis with local information computed on demand during
the taint propagation. This removes the need of recomputing
these data structures when the target changes. For the
data-
ow facts as such, their tool Andromeda computes the
transitive closure of all facts in
uenced by the code change,
removes them, and then starts a recomputation. While
the latter is quite close to Reviser's clear-and-propagate
approach, Reviser only deletes jump functions on demand
and thus can abort earlier if it detects that a jump function
would be recreated in exactly the same way, i.e., it does not
delete functions that might not change at all just because
they are in the change's transitive dependence set.
Eichberg et al. [6] use incremental tabled evaluation to
update analysis results deduced in a logic language. This
approach comes with the cost of having to translate the pro-
gram graph to Prolog facts and depends on the performance
of the underlying general-purpose reasoning engine.
空白的是data-flow