[llvm-dev] [RFC] Changing `lit` default output
Reid Kleckner via llvm-dev
llvm-dev at lists.llvm.org
Tue May 31 17:01:03 PDT 2016
As long as the individual commands are still copy-pastable into both cmd
and bash, I'm happy. I think practice this means the output should look
more like this:
// RUN: %clang_cc1 %s -verify
$ "C:\\src\\llvm\\build\\bin\\clang.exe" -cc1 "C:\\....\\test.cpp" -verify
Essentially every program used in the LLVM test suite understands forward
slashes, so we could rewrite the path to test.cpp to use forward slashes,
but I think cmd wants to see backslashes in argv0. Then our output would
$ "C:\\src\\llvm\\build\\bin\\clang.exe" -cc1 C:/src/llvm/.../test.cpp
On Tue, May 31, 2016 at 4:53 PM, Daniel Dunbar via llvm-dev <
llvm-dev at lists.llvm.org> wrote:
> (+ the real cfe-dev)
> On Tue, May 31, 2016 at 4:48 PM, Daniel Dunbar <daniel at zuster.org> wrote:
>> Hi all,
>> I have some patches up in:
>> to make `lit` output more readable when commands fail. I'd like to get a
>> little more feedback on this before landing, since it impacts how test
>> failures appear when using the internal shell script runner (which is not
>> the default so this will currently mostly impact Windows or test frameworks
>> that intentionally set this behavior).
>> These patches change the output of a script like `true && echo hi &&
>> false` to something like:
>> $ true
>> $ echo hi
>> $ false
>> note: command had no output on stdout or stderr
>> error: command failed with exit status 1
>> instead of the old:
>> Command 0: "true"
>> Command 0 Result: 0
>> Command 0 Output:
>> Command 0 Stderr:
>> Command 1: "echo" "hi"
>> Command 1 Result: 0
>> Command 1 Output:
>> Command 1 Stderr:
>> Command 2: "wc" "missing-file"
>> Command 2 Result: 1
>> Command 2 Output:
>> Command 2 Stderr:
>> which I think is significantly more readable.
>> One change in the patches is that it will also start including, inline,
>> the output of files which were redirected. This solves a common problem
>> where a command fails when it wasn't expected to, and its output was
>> captured. In such cases the test log just showed the command failure, but
>> since the output was redirected you couldn't tell from the log why the
>> common failed.
>> Please let me know if you have any objections to the new format.
>> - Daniel
> LLVM Developers mailing list
> llvm-dev at lists.llvm.org
-------------- next part --------------
An HTML attachment was scrubbed...
More information about the llvm-dev