<html>
<head>
<base href="http://llvm.org/bugs/" />
</head>
<body><span class="vcard"><a class="email" href="mailto:renato.golin@linaro.org" title="Renato Golin <renato.golin@linaro.org>"> <span class="fn">Renato Golin</span></a>
</span> changed
<a class="bz_bug_link
bz_status_RESOLVED bz_closed"
title="RESOLVED WONTFIX - ARM: inline asm does not support pre-UAL conditional LDRH etc"
href="http://llvm.org/bugs/show_bug.cgi?id=11855">bug 11855</a>
<br>
<table border="1" cellspacing="0" cellpadding="8">
<tr>
<th>What</th>
<th>Removed</th>
<th>Added</th>
</tr>
<tr>
<td style="text-align:right;">Status</td>
<td>NEW
</td>
<td>RESOLVED
</td>
</tr>
<tr>
<td style="text-align:right;">Resolution</td>
<td>---
</td>
<td>WONTFIX
</td>
</tr></table>
<p>
<div>
<b><a class="bz_bug_link
bz_status_RESOLVED bz_closed"
title="RESOLVED WONTFIX - ARM: inline asm does not support pre-UAL conditional LDRH etc"
href="http://llvm.org/bugs/show_bug.cgi?id=11855#c10">Comment # 10</a>
on <a class="bz_bug_link
bz_status_RESOLVED bz_closed"
title="RESOLVED WONTFIX - ARM: inline asm does not support pre-UAL conditional LDRH etc"
href="http://llvm.org/bugs/show_bug.cgi?id=11855">bug 11855</a>
from <span class="vcard"><a class="email" href="mailto:renato.golin@linaro.org" title="Renato Golin <renato.golin@linaro.org>"> <span class="fn">Renato Golin</span></a>
</span></b>
<pre>Mans, others,
I had extensive talks with key ARM people spanning across multiple compiler
toolchains, companies and countries (some of them replied to this post), and I
have reached the conclusion that it's not profitable, not even desirable, to
implement non-UAL assembly directives in LLVM. It is a unanimous opinion that
there is no future for non-UAL assembly.
Other toolchains already had pre-UAL support, and maintaining them costs almost
nothing, so there is no incentive to remove the feature. But implementing them
in LLVM makes no sense. GCC's lack of support for UAL is troubling and should
be dealt with on its own, without affecting the direction of LLVM. The cost
associated with implementing it in LLVM is big and the benefit is small. We'd
rather focus our attention to other matters, including UAL assembly support.
LLVM-based compilers, especially Clang, focus on being standards compliant, not
GNU compliant, and that's an important tool for the rest of the community, that
uses the fact that we have now two powerful open source C/C++ compilers, with
which they can test and find more errors on their programs. If Clang follows
GCC on every step, the community at large would lose that ability, and there
would be little point for LLVM.
Although we understand that legacy code cannot be compiled with Clang, this is
not an entirely bad thing, as we currently have two solutions for this:
1. Use GAS, GCC etc, as little as possible, on the restricted subset where
Clang+LLVM can't compile
The same is true for any other legacy compiler, including the Intel and the ARM
compilers. There's nothing new here, and as an example of their quality,
projects that rely on their idiosyncrasies are summarily abandoned when they
hit the open source world.
2. Work with the developers to upgrade their code, use new libraries, write new
code.
It's a very noble activity that of updating your software. Machines are not
what they used to be, algorithms get better, compilers get better, standards
get better. The more we encourage people to stick to standards, the better.
Especially since the number of reasons not to is getting smaller every decade.
I'm closing this bug as "wontfix", and I hope not to offend anyone with this,
but there needs to be a line drawn somewhere. We will continue to accept the
removal of pre-UAL syntax from LLVM as a means to get the integrated assembler
production quality for ARM as soon as possible.
Feel free to reference this text on other/future bugs about non-UAL syntax.</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>