<html>
<head>
<base href="https://bugs.llvm.org/">
</head>
<body><table border="1" cellspacing="0" cellpadding="8">
<tr>
<th>Bug ID</th>
<td><a class="bz_bug_link
bz_status_NEW "
title="NEW - AST should explain the implicitness of the casts better"
href="https://bugs.llvm.org/show_bug.cgi?id=38166">38166</a>
</td>
</tr>
<tr>
<th>Summary</th>
<td>AST should explain the implicitness of the casts better
</td>
</tr>
<tr>
<th>Product</th>
<td>clang
</td>
</tr>
<tr>
<th>Version</th>
<td>trunk
</td>
</tr>
<tr>
<th>Hardware</th>
<td>PC
</td>
</tr>
<tr>
<th>OS</th>
<td>Linux
</td>
</tr>
<tr>
<th>Status</th>
<td>NEW
</td>
</tr>
<tr>
<th>Severity</th>
<td>enhancement
</td>
</tr>
<tr>
<th>Priority</th>
<td>P
</td>
</tr>
<tr>
<th>Component</th>
<td>Frontend
</td>
</tr>
<tr>
<th>Assignee</th>
<td>unassignedclangbugs@nondot.org
</td>
</tr>
<tr>
<th>Reporter</th>
<td>lebedev.ri@gmail.com
</td>
</tr>
<tr>
<th>CC</th>
<td>llvm-bugs@lists.llvm.org
</td>
</tr></table>
<p>
<div>
<pre>Related to #38017.
We can't actually distinguish whether the implicit cast is implicit in source
code, or is part of explicit cast.
We can't use ASTContext::getParents() (partly performance, partly semantics).
But we also sometimes don't visit the ExplicitCastExpr.
<zygoloid> CodeGenFunction::EmitMaterializeTemporaryExpr calls
skipRValueSubobjectAdjustments, which steps over the CK_NoOp cast
<zygoloid> OK, so we could fix that, or we could add some state to the CastExpr
so you can tell whether it's an "implicitly doing bad things" cast kind
<zygoloid> or maybe a CastKind; we could use CK_UncheckedIntegralCast for
"integral cast where the user explicitly asked for it" versus CK_IntegralCast
for "integral cast that was performed implicitly"
<zygoloid> i don't think "is part of explicit cast" is quite right; given
"struct X { X(char); }", X(some_int) should still be checked
<zygoloid> but perhaps "enum E : char {}; E(some_int)" is more borderline
<LebedevRI> which solution does sound best to you? the ast one, or something
involving skipRValueSubobjectAdjustments() ? /me needs to know how to title the
bug
<zygoloid> I think it makes sense to make skipRValueSubobjectAdjustments skip
the minimum amount of stuff necessary to get to the expression constructing the
temporary
<zygoloid> but I'd be OK with having a new cast kind for this, because in the
generalized Clang model (including UBSan) the cast has different semantics</pre>
</div>
</p>
<hr>
<span>You are receiving this mail because:</span>
<ul>
<li>You are on the CC list for the bug.</li>
</ul>
</body>
</html>