<html>
<head>
<meta content="text/html; charset=windows-1252"
http-equiv="Content-Type">
</head>
<body bgcolor="#FFFFFF" text="#000000">
I'd be very happy to see this turned on as a warning; I'd be very
reluctant to see it turned on as a error. During a lot of code
cleanups, the intermediate state of "deleted, but dead args still
passed in" is a very useful intermediate step. It greatly
simplifies review to be able to separate the meat of the semantic
change from the "just deleting a bunch of dead args" patch. I'd be
very very reluctant to create a situation where the patch with the
semantic change but without the argument cleanup won't pass all the
build bots. <br>
<br>
Philip<br>
<br>
(1 inline comment below as well)<br>
<br>
<div class="moz-cite-prefix">On 07/18/2015 02:48 AM, Chandler
Carruth wrote:<br>
</div>
<blockquote
cite="mid:CAGCO0Kiquf0As+CsQjt+BbToDCu1Mj=8-UCcY6abhzVUTmNSOQ@mail.gmail.com"
type="cite">
<div dir="ltr">LLVM and Clang both have lots of objects that are
passed through many different API boundaries. Things like
AliasAnalysis in LLVM or Sema in Clang get threaded all over the
place.
<div><br>
</div>
<div>Over times, refactoring can often cause the parameters (or
local variables, or member variables, etc) to become dead. If
we notice this, we can often un-thread the interface through
our APIs, sometimes even reducing coupling, etc.</div>
<div><br>
</div>
<div>But currently -Wunused-parameter is hard disabled. I've not
checked to see if there are any other disabling of -Wunused-*
variants, but I'd like to move the project toward firmly
enabling them and being clean with them.</div>
<div><br>
</div>
<div>The technique I'd like to use is leaving out names of
unused parameters, using a /*CommentedName*/ if there is a
useful name, otherwise just omitting the name.</div>
</div>
</blockquote>
Sorry, could you rephrase? Not sure what you're getting at here.<br>
<blockquote
cite="mid:CAGCO0Kiquf0As+CsQjt+BbToDCu1Mj=8-UCcY6abhzVUTmNSOQ@mail.gmail.com"
type="cite">
<div dir="ltr">
<div><br>
</div>
<div>All of this will require a reasonable amount of cleanup
across the projects which I'm happy to do prior to flipping
defaults around. Thoughts? Any concerns or objections?</div>
<div><br>
</div>
<div>(I'll ask the same question and ofter to either enact the
cleanups or toggle the warning(s) back off for each of the
less intertwined subprojects like LLD, LLDB, Polly, etc.)</div>
<div><br>
</div>
<div>-Chandler</div>
</div>
<br>
<fieldset class="mimeAttachmentHeader"></fieldset>
<br>
<pre wrap="">_______________________________________________
LLVM Developers mailing list
<a class="moz-txt-link-abbreviated" href="mailto:LLVMdev@cs.uiuc.edu">LLVMdev@cs.uiuc.edu</a> <a class="moz-txt-link-freetext" href="http://llvm.cs.uiuc.edu">http://llvm.cs.uiuc.edu</a>
<a class="moz-txt-link-freetext" href="http://lists.cs.uiuc.edu/mailman/listinfo/llvmdev">http://lists.cs.uiuc.edu/mailman/listinfo/llvmdev</a>
</pre>
</blockquote>
<br>
</body>
</html>