[llvm-bugs] [Bug 52297] New: interesting difference in do_get behavior between libc++ and other libraries.
via llvm-bugs
llvm-bugs at lists.llvm.org
Mon Oct 25 12:47:41 PDT 2021
https://bugs.llvm.org/show_bug.cgi?id=52297
Bug ID: 52297
Summary: interesting difference in do_get behavior between
libc++ and other libraries.
Product: libc++
Version: 11.0
Hardware: PC
OS: All
Status: NEW
Severity: enhancement
Priority: P
Component: Standards Issues
Assignee: unassignedclangbugs at nondot.org
Reporter: edward.vogel at hpe.com
CC: llvm-bugs at lists.llvm.org, mclow.lists at gmail.com
Created attachment 25391
--> https://bugs.llvm.org/attachment.cgi?id=25391&action=edit
Small program that shows the issue.
The attached small program shows a difference in the behavior of libc++ do_get
when a class overrides the standard do_get. Here an overriding do_get for
bool types calls the base class do_get. With libc++ that call will call the
long do_get viritual function in the overriding class. Other libraries (gnu,
Visual Studio) will call the base class long do_get.
In reading the Standard, it's not clear to me who is correct. This is why I've
posted this as a standards issue.
Here is a cut/paste from a Cygwin session:
VogelEd at XLB3502Q4E ~/temp
$ clang++ -o a.out t1.cpp
VogelEd at XLB3502Q4E ~/temp
$ ./a.out
In bool do_get
Returning from bool do_get
VogelEd at XLB3502Q4E ~/temp
$ clang++ -o a.out -stdlib=libc++ t1.cpp
VogelEd at XLB3502Q4E ~/temp
$ ./a.out
In bool do_get
In long do_get
Returning from long do_get
Returning from bool do_get
The attached reproducer is a greatly reduced and modified version of one of
the runtime library tests I use to test one of our HPE runtimes.
Probably not a big deal, but this difference in behavior could cause customer
confusion.
Thank you for your time,
Ed Vogel
--
You are receiving this mail because:
You are on the CC list for the bug.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/llvm-bugs/attachments/20211025/917d68d0/attachment.html>
More information about the llvm-bugs
mailing list