patch: add mangling for attribute enable_if

[Richard Smith]

Your lambda doesn't capture anything -- maybe use [] rather than [&]?
(Maybe we should warn on this.)

Maybe use a reverse_iterator loop over FD->getAttrs, and skip the
non-EnableIfAttrs. That'd remove the need to make a copy of the vector and
to reverse it. (Do we have llvm::reversed yet? If so, 'for (Attr *A :
reversed(FD->getAttrs()))' seems like a nice way to write this.)


On Sun, Mar 9, 2014 at 5:02 PM, Nick Lewycky <nicholas at mxc.ca> wrote:

> On 03/09/2014 11:01 AM, David Majnemer wrote:
>
>> Your std::remove_if would be more concise if it used a lambda instead
>> of IsNotEnableIfAttr.
>>
>
> Done. Also added another test case to the mangling tests.
>
> Nick
>
>  On Sun Mar 09 2014 at 10:30:45 AM, Nick Lewycky <nicholas at mxc.ca
>> <mailto:nicholas at mxc.ca>> wrote:
>>
>>     The attached mangle-enable_if-1.patch adds a mangling for
>>     __attribute__((enable_if(expr, string-literal))) to clang.
>>
>>     demangle-enable_if-1.patch adds support to libcxxabi's demangler. This
>>     patch I'm not very confident in. libcxxabi's cxa_demangle lacks
>> comments
>>     and assertions, leaving its design criteria a mystery. I think the
>>     functions return 'first' in case of error. I added a vector<> inside
>> the
>>     demangler, I don't know whether that's OK because it means doing
>>     allocation. I don't know what the two strings in the pair in db.names
>>     are for, but the first one appears to be the demangling so I put it
>>     there. I don't understand what the members in 'db' are for, since they
>>     aren't commented (db.tag_templates?). The code is cargo culted from
>>     parse_template_args and simplified down by making wild assumptions
>>     (db.tag_templates is always false!) and constant folding.
>>
>>     Also the demangler seems pretty buggy. It wraps expressions in extra
>>     parentheses which don't correspond to pi .. E expressions (if you were
>>     to remangle it, you'd get the extra pi .. E). A mangling for
>> "&function"
>>     ends up demangling "&(function())" which has different semantic
>> meaning.
>>        I'm assuming these problems are pre-existing. While "_Z3foo"
>>     demangles
>>     to "foo" and "_Z3foov" demangles to "foo()", the same things with an
>>     attribute demangle to include parens. For example, "_Z3fooUa3bar" and
>>     "_Z3fooUa3barv" both demangle to "foo() __attribute__((bar))", never
>>     "foo __attribute__((bar))".
>>
>>     Finally, I've never even compiled the change to the tests. I have
>> tested
>>     those exact manglings and seen what cxa_demangle does to them, but it
>>     was easier to write a standalone tool than to deal with "testit".
>>
>>     Please review!
>>
>>     Nick
>>     _________________________________________________
>>     cfe-commits mailing list
>>     cfe-commits at cs.uiuc.edu <mailto:cfe-commits at cs.uiuc.edu>
>>     http://lists.cs.uiuc.edu/__mailman/listinfo/cfe-commits
>>     <http://lists.cs.uiuc.edu/mailman/listinfo/cfe-commits>
>>
>>
>
> _______________________________________________
> cfe-commits mailing list
> cfe-commits at cs.uiuc.edu
> http://lists.cs.uiuc.edu/mailman/listinfo/cfe-commits
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/cfe-commits/attachments/20140309/f33de249/attachment.html>