<div class="gmail_quote">On Sat, Jul 30, 2011 at 8:57 AM, Howard Hinnant <span dir="ltr"><<a href="mailto:hhinnant@apple.com">hhinnant@apple.com</a>></span> wrote:<br><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex;">
<div id=":8dx">It is not clear to me that the addition of ext/slist is a path we want to take with libc++. It bloats the library. It requires debugging, maintenance, and unit tests. There exists a C++11 replacement that is well specified and well tested.<br>
<br>
We don't need another C++03 std::lib. Plenty of those already exists. libc++ was envisioned to be a C++0x library. There are tons of old extensions libc++ doesn't have. Are we going to implement all of them? Even the ones that have C++11 replacements? Even the ones that introduce their own set of problems? Where do we draw the line?<br>
</div></blockquote><div><br></div><div>These bring up all very good points. I'll save most of the discussion for a proper thread on cfe-dev on the subject, but I don't think there is any disagreement that we don't want yet another C++03 lib.</div>
<div><br></div><div>I've reverted the patches introducing slist in r136577. I think it might be a useful thing to include specifically to aid in migration from other standard libraries, but clearly that's the discussion that needs to be had on the proper list.</div>
<div><br></div><div>If we ever do want slist, it will need proper tests, and a more bullet proof implementation *specifically because* it is adding more maintenance cost to the library. Without those, there's no question at all.</div>
</div>