[Patch][LNT] fix python error in compile.py when runN failed

[Daniel Dunbar]

On Wed, Jun 25, 2014 at 8:30 AM, Chris Matthews <chris4000 at gmail.com> wrote:

> I think the message failed to parse output is misleading.
>
> The problem that I have experienced several times is that no output is
> printed, so all I get is a "compile.py:71: error: failed to parse output: “
> and nothing else, which then makes it look like LNT had a problem, not the
> subprocess.
>
> For instance, this is a snippet from a recent log:
>
> compile.py:71: error: failed to parse output:
> compile.py:74: error: stderr isn't empty: runN: ERROR: child returned
> error: 65.
> compile.py:78: error: res != 0: 1
>
>
> Not super helpful.
>

That makes sense, and definitely not super helpful.

I put in a slightly different patch to make this more readable here:
  http://llvm.org/viewvc/llvm-project?view=revision&revision=211719

Steven, can you verify that resolves the problem you were seeing?

Now you get output like:
----
2014-06-25 10:15:27: ERROR: command failed with stderr:
--
/Users/ddunbar/test-suite-externals/lnt-compile-suite-src/intel/single-file/403.gcc/combine.c:12778:2:
error: foo
#error foo
 ^
1 error generated.
runN: ERROR: child returned error: 1.
--
----
or
----
2014-06-25 10:15:50: ERROR: command had unexpected output on stderr:
--
/Users/ddunbar/test-suite-externals/lnt-compile-suite-src/intel/single-file/403.gcc/combine.c:12778:2:
warning: foo [-W#warnings]
#warning foo
 ^
1 warning generated.
--
----

 - Daniel


>
> On Jun 25, 2014, at 7:52 AM, Steven Wu <stevenwu at apple.com> wrote:
>
> I agree that the original code works but it is doing more work than
> necessary and throw ugly python errors. This is a performance & cosmetic
> fix.
>
> Steven
>
> On Jun 24, 2014, at 10:27 AM, Daniel Dunbar <daniel at zuster.org> wrote:
>
> It isn't clear to me this is the right patch. Where are you seeing "python
> errors" being dumped during "runtest compile"?
>
> If I inject an error into one of the source files, what I see is this:
> --
> $ lnt runtest compile --sandbox $(timestamp) --cc $(which clang)
> --test-externals ~/test-suite-externals --test-subdir
> lnt-compile-suite-src/intel --test-filter=compile/403.gcc/combine.c/irgen
> --flags-to-test="-O0"
> compile.py:779: note: inferred C++ compiler under test as:
> '/usr/bin/clang++'
> 2014-06-24 17:25:04: creating sandbox: '2014-06-24_10-25'
> compile.py:893: note: inferred run order to be: '600.0.34'
> 2014-06-24 10:25:04: INFO: Loading config file:
> /Users/ddunbar/test-suite-externals/lnt-compile-suite-src/intel/project_list.json
> 2014-06-24 10:25:04: INFO: run started
> 2014-06-24 10:25:04: INFO: using CC: '/usr/bin/clang'
> 2014-06-24 10:25:04: INFO: using CXX: '/usr/bin/clang++'
> compile.py:71: error: failed to parse output:
>
> compile.py:74: error: stderr isn't empty:
> /Users/ddunbar/test-suite-externals/lnt-compile-suite-src/intel/single-file/403.gcc/combine.c:12776:1:
> error: unknown type name 'xxxx'
> xxxx
> ^
> /Users/ddunbar/test-suite-externals/lnt-compile-suite-src/intel/single-file/403.gcc/combine.c:12777:5:
> error: expected ';' after top level declarator
> xxxx
>     ^
>     ;
> 2 errors generated.
> runN: ERROR: child returned error: 1.
>
>
> compile.py:78: error: res != 0: 1
> --
> which isn't pretty, but it works.
>
>  - Daniel
>
>
>
> On Tue, Jun 24, 2014 at 9:34 AM, Steven Wu <stevenwu at apple.com> wrote:
>
>> Sorry I forgot to attach the patch in the previous email. Now attached.
>>
>>
>> > On Jun 23, 2014, at 6:06 PM, Steven Wu <stevenwu at apple.com> wrote:
>> >
>> > LNT currently throws python errors when runN failed in compile.py. It
>> throws error messages about it cannot parse stdout, stderr is not empty and
>> runN does not return 0.
>> > Ideally, the check after runN should performed in the opposite order.
>> stdout shouldn’t be parsed if runN does not return 0. Since it is expected
>> that runN can fail during the test, g_log.error should be used to print the
>> error instead of throw python error messages.
>> >
>> > Please commit for me after review.
>>
>>
>> _______________________________________________
>> llvm-commits mailing list
>> llvm-commits at cs.uiuc.edu
>> http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits
>>
>>
>
> _______________________________________________
> llvm-commits mailing list
> llvm-commits at cs.uiuc.edu
> http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits
>
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/llvm-commits/attachments/20140625/745345a8/attachment.html>