rebase to head.

rebase to head.

address comment.

fix testcase.

format again.

format code.

address comment.

address comment.

Move some desugared type to Definition.

rename TTD to TAT.

handle TypeAliasTemplateDecl.

1. rename CanType to DesugaredTy.
2. fix some bug in compare PrintedType.

fix some bug.

address comment.

Related discussions: D72498

format code and add testcase for template.

add testcase

In D82548#2124546, @MaskRay wrote:

Hi, your git commit contains extra Phabricator tags. You can drop Reviewers: Subscribers: Tags: and the text Summary: from the git commit with the following script:

arcfilter () {
        arc amend
        git log -1 --pretty=%B | awk '/Reviewers:|Subscribers:/{p=1} /Reviewed By:|Differential Revision:/{p=0} !p && !/^Summary:$/ {sub(/^Summary: /,"");print}' | git commit --amend --date=now -F -
}

Reviewed By: is considered important by some people. Please keep the tag. (--date=now is my personal preference (author dates are usually not useful. Using committer dates can make log almost monotonic in time))

https://reviews.llvm.org/D80978 contains a git pre-push hook to automate this.

Add the missing operator!=.

Address comments.

Rebase

Rebase to head.

Rebase to head.

Rebase to head.

Address comment.

format code.

Address comment.
I will add using $alias$ = $type$; in a separate patch.

If this patch looks good, please land it for me.

address comment.
Also tweak code completion behavior for using.

I think the character "🡺" should be avoided, as it may not display properly in some environments.

In D72498#1816339, @sammccall wrote:

In D72498#1816244, @ilya-biryukov wrote:

In D72498#1815500, @lh123 wrote:

• hover over the front , you'll see "instance-method front → std::vector<int, class std::allocator<int> >::reference".
• hover over the push_back, you'll see "std::vector<int, class std::allocator<int> >::value_type && __x".

These look terrible and are the great examples where showing canonical types results in better output than canonical types.
I wonder why we add std::vector<int, class std::allocator<int>>:: in the first place, I believe the standard library uses value_type in the declaration. Showing value_type is not great, but at least that doesn't uglify what was written in the code in the first place.
FWIW, I think the perfect output in those cases would be int (aka value_type)

Indeed. Another illustrative example, the return type of vector<int64_t>::at() - we'd probably want int64& here, rather than vector<...>::reference or unsigned long long/unsigned long depending on platform.

In D72500#1813975, @sammccall wrote:

I think we should avoid triggering for literals. Maybe some exceptions, but a hover saying that 0 is an int with value 0 seems silly.

In D72498#1813902, @ilya-biryukov wrote:

Could it be the case that we want to show the canonical types (i.e. without all syntax sugar)?
Maybe we want both the normal type and the canonical type?

what do you think about unwrapping decltype only when it's a return value (optional: of a function whose leading return type is auto) to narrowly catch this idiom?

In D72498#1814008, @sammccall wrote:

I think i'm also comfortable with marking the linked bug as wontfix.

@sammccall Could you commit this, I don't have commit access.

Address comments.

In D70222#1763553, @kadircet wrote:

Thanks a lot for working on this patch, LGTM!

again please make sure the changes are clang-formatted.

I'm sure the code has been formatted.

fix typo:
"Could not convert UTF16 To UTF8" -> "Could not convert UTF16 to UTF8"

address comments.

address comments.

address comments.

fixes no matching constructor for initialization of llvm::StringError.

address comments.

address comments.

fix bug.

Sorry, upload the wrong patch.

address comments.