labath added a comment.
In https://reviews.llvm.org/D49740#1188079, @teemperor wrote:
> > CompletionRequest - this sounds like it could go next to the command
> > interpreter
>
> Yeah, makes sense. Even though Utility classes then can either not offer
> completion methods (currently only ArchSpec is doing that) or work around
> that with a forward decl. But I think that's not a big issue.
I am not too bothered by that either, but I think that for the thing that the
ArchSpec function is doing, passing it the entire CompletionRequest is
overkill. Ideally, I'd set things up such that ArchSpec and friends don't have
any knowledge of "completion" -- they would just offer an interface for
accessing all possible values(*), and then completion could be built on top of
that. This way the data source can be independent of the actual algorithm
computing the completions.
(*) The interface could be as simple as an iterator over all ArchSpec string
values. However, this may not be good enough in terms of performance (it
wouldn't matter for ArchSpec, but it might for some other classes), in which
case we might want to do something slightly more fancy
(`GiveMeAllValuesStartingWith("foo")`).
https://reviews.llvm.org/D49740
_______________________________________________
lldb-commits mailing list
lldb-commits@lists.llvm.org
http://lists.llvm.org/cgi-bin/mailman/listinfo/lldb-commits
