[Lldb-commits] [PATCH] D48351: Move dumping code out of RegisterValue class

Zachary Turner via lldb-commits lldb-commits at lists.llvm.org
Fri Jul 20 08:04:01 PDT 2018

We only use dumping from lldb-test, from inside the debugger, or from the
top level command interpreter. All of those things necessarily depend on
everything anyway.
On Fri, Jul 20, 2018 at 8:02 AM Pavel Labath <labath at google.com> wrote:

> Well, if it depends on everything, then it can only used from places
> that also depend on everything. Right now I don't think it matters now
> (though it will create a new Dump <-> "everything calling Dump"
> cycle). However, I can imagine this might cause some tricky situations
> later on when we try to clean up dependencies between the high level
> classes.
> On Fri, 20 Jul 2018 at 15:50, Zachary Turner via Phabricator via
> lldb-commits <lldb-commits at lists.llvm.org> wrote:
> >
> > zturner added a subscriber: labath.
> > zturner added a comment.
> >
> > I had previously thought of making a top level project called Dump that
> > depends on everything. Also makes it very obvious where all the dumpers
> > are. It can have overloaded functions called lldb_private::dump(T&) for
> > every value of T, then no matter what type you have, all you have to do
> is
> > call dump(T). Kills all birds with one stone
> Also remember that our dump functions have various (and
> non-homogeneous) arguments to tweak the dump format. That doesn't mean
> this couldn't work, just that the calls will look more like dump(T,
> Stream, format_arg_1, format_arg_2, ... format_arg_n).
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/lldb-commits/attachments/20180720/3d2efd75/attachment.html>

More information about the lldb-commits mailing list