[Lldb-commits] [lldb] [lldb/Target] Unify frame provider descriptor and chain IDs (PR #190712)
Jonas Devlieghere via lldb-commits
lldb-commits at lists.llvm.org
Tue Apr 7 09:56:52 PDT 2026
================
@@ -258,6 +258,7 @@ static constexpr CommandObject::ArgumentTableEntry g_argument_table[] = {
{ lldb::eArgTypeFilename, "filename", lldb::eDiskFileCompletion, {}, { nullptr, false }, "The name of a file (can include path)." },
{ lldb::eArgTypeFormat, "format", lldb::CompletionType::eNoCompletion, {}, { FormatHelpTextCallback, true }, nullptr },
{ lldb::eArgTypeFrameIndex, "frame-index", lldb::eFrameIndexCompletion, {}, { nullptr, false }, "Index into a thread's list of frames." },
+ { lldb::eArgTypeFrameProviderIDRange, "frame-provider-id-range", lldb::CompletionType::eNoCompletion, {}, { nullptr, false }, "A single frame provider ID, a range of IDs (e.g., '0', '0-2', '0 to 2'), or '*'/'all' to show every provider. ID 0 is the base unwinder, 1+ are synthetic providers." },
----------------
JDevlieghere wrote:
> A single frame provider ID (e.g. '0'), a range of IDs (e.g. '0-2'), or 'all' to show every provider. ID 0 is the base unwinder, 1+ are synthetic providers.
I'm skeptical of supporting too many variants. It increases complexity (both in the implementation and the test) and makes it confusing to users. I'd argue it's often better to stick to one thing that's consistent and opinionated as they're easier to remember.
https://github.com/llvm/llvm-project/pull/190712
More information about the lldb-commits
mailing list