ioeric added inline comments.
================
Comment at: clangd/Cancellation.h:96
+/// checks using it to avoid extra lookups in the Context.
+class CancellationToken {
+public:
----------------
ilya-biryukov wrote:
> ioeric wrote:
> > As chatted offline, I have questions about the motivation of the
> > `CancellationToken` class. Intuitively, it seems to me that this can be
> > merged into `TaskHandle`, and we can simply stash the `TaskHandle` instead
> > of a token into the context. There would be fewer states to maintain, and
> > the design would be a bit more straightforward. I might be missing
> > context/concerns here, so I'd like to hear what you and Ilya think.
> > @ilya-biryukov
> I am for splitting cancellation checks from cancellation triggers.
> The processing code only needs to check if it was cancelled and exposing the
> `cancel()` does not add any useful functionality, merely ways to misuse it.
Couldn't we prevent this by passing only const handle references to users who
are not expect to call `cancel()`?
Repository:
rCTE Clang Tools Extra
https://reviews.llvm.org/D50502
_______________________________________________
cfe-commits mailing list
cfe-commits@lists.llvm.org
http://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
