[cfe-dev] clang-format turning tests into no-ops

[Sean Silva]

On Fri, Nov 8, 2013 at 12:49 AM, Alp Toker <alp at nuanti.com> wrote:

>  Hi Daniel,
>
> Here's a good one for you..
>
> I've recently been fixing a whole lot of lit tests in the LLVM / clang
> suites that don't check anything.
>
> It was a mystery why people keep checking in tests like this:
>
> // RUN: clang-check "%s" -- -target x86_64-apple-darwin10 -fasm-blocks
> 2>&1 |
> // FileCheck %s
>
> The first line will run successfully, and the second doesn't begin with
> RUN so the test passes silently. Not good.
>
> Then it dawned on me -- contributors must be running clang-format before
> committing, causing their RUN lines to get split up at the line boundary.
>
> I know this isn't strictly a clang-format bug, but it's time consuming to
> manually audit and find no-op tests like this after the fact and the
> failure mode is insidious because these tests will always silently pass.
>

Although it doesn't eliminate the hassle of having to manually fix this, it
seems like at least one issue worth fixing in its own right is the fact
that RUN lines ending with a | are silently accepted by our test
infrastructure.

Also, FWIW, it seems like clang-format's comment reflowing turns out to do
the wrong thing in a ton of (most?) scenarios (for example, it will wrap
the last word onto the next line, leaving a bunch of spurious "one word"
lines), to the point that I wish there were an option to tell clang-format
to not mess with comments at all.

-- Sean Silva


>
> I was wondering if you can think of a way to exclude the test suites from
> clang-format and clang-format-diff runs. Guessing this could be tricky
> given that there are several different ways to run clang-format but
> something needs to be done..
>
> Alp.
>
> -- http://www.nuanti.com
> the browser experts
>
>
> _______________________________________________
> cfe-dev mailing list
> cfe-dev at cs.uiuc.edu
> http://lists.cs.uiuc.edu/mailman/listinfo/cfe-dev
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/cfe-dev/attachments/20131108/5f902add/attachment.html>