tfiala closed this revision.
tfiala added a comment.
Closed via commit:
r24
http://reviews.llvm.org/D19215
___
lldb-commits mailing list
lldb-commits@lists.llvm.org
http://lists.llvm.org/cgi-bin/mailman/listinfo/lldb-commits
tfiala updated this revision to Diff 54102.
tfiala added a comment.
Final change: this change:
1. has the lldbinline.py test intercept its __file parameter and convert from
.pyc to .py before storing to the test.test_filename attribute.
2. has the test event creation mechanism and test_event usa
tfiala added a comment.
Side note:
Okay I think the mystery of this was caught by Jim. When we started accepting
usage of .pycs again, we're allowing the python runtime to select using the pyc
if available and not older than the .py file. This makes it possible to get a
test file that is som
tfiala added a comment.
> The __file__ for lldbinline.py-based tests is getting set to .pyc via the
> unittest2 loader. This callstack comes from inserting this code into
> lldbinline.py:
>
> diff --git a/packages/Python/lldbsuite/test/lldbinline.py
> b/packages/Python/lldbsuite/test/lldb
tfiala added a comment.
In http://reviews.llvm.org/D19215#404143, @zturner wrote:
> It seems very strange to me to be changing a .pyc filename to a .py
> filename. I think we should try to understand where the .pyc filename is
> coming from to begin with, and this is just masking the real erro
They make the test suite run faster. when a file wasn't changed.
On Mon, Apr 18, 2016 at 11:34 AM Jim Ingham wrote:
> That I don't know. But we really shouldn't be generating the .pyc files,
> they just litter the test directories and I don't think they are much help.
>
> Jim
>
> > On Apr 18, 2
(not sure how much though, as I haven't actually measured it)
On Mon, Apr 18, 2016 at 11:40 AM Zachary Turner wrote:
> They make the test suite run faster. when a file wasn't changed.
>
> On Mon, Apr 18, 2016 at 11:34 AM Jim Ingham wrote:
>
>> That I don't know. But we really shouldn't be gene
That I don't know. But we really shouldn't be generating the .pyc files, they
just litter the test directories and I don't think they are much help.
Jim
> On Apr 18, 2016, at 11:29 AM, Zachary Turner wrote:
>
> That's normal, but why would that cause an issue for the test suite? What is
> l
That's normal, but why would that cause an issue for the test suite? What
is leading to it thinking that it should execute a .pyc file in the first
place?
On Mon, Apr 18, 2016 at 11:25 AM Jim Ingham wrote:
> IIRC, python will generate byte-compiled ".pyc" versions of the ".py"
> files automatic
IIRC, python will generate byte-compiled ".pyc" versions of the ".py" files
automatically (and helpfully leave them in the CWD...) We turned that off for
the testsuite through some Python magic I don't think I ever knew, but if I did
I've certainly forgotten it... Anyway, sounds like that got
It seems very strange to me to be changing a .pyc filename to a .py
filename. I think we should try to understand where the .pyc filename is
coming from to begin with, and this is just masking the real error.
On Mon, Apr 18, 2016 at 10:10 AM Pavel Labath via lldb-commits <
lldb-commits@lists.llvm
labath accepted this revision.
labath added a comment.
This revision is now accepted and ready to land.
Ok, makes sense then, thanks for making sure it's necessary.
I don't quite understand how the string makes it's way there though, as the
only place setting this field in the entire repository
tfiala updated this revision to Diff 54079.
tfiala added a comment.
Whoops - don't include the 'raise' call.
http://reviews.llvm.org/D19215
Files:
packages/Python/lldbsuite/test/result_formatter.py
Index: packages/Python/lldbsuite/test/result_formatter.py
tfiala updated this revision to Diff 54078.
tfiala added a comment.
Change inspect.getfile() => inspect.getsourcefile().
http://reviews.llvm.org/D19215
Files:
packages/Python/lldbsuite/test/result_formatter.py
Index: packages/Python/lldbsuite/test/result_formatter.py
tfiala added a comment.
> I was able to replicate just prior to applying the change, but I'll double
> check.
Yeah that change didn't cover all the cases. The test_filename that comes from
this call stack (in startTest), originating within the unittest2 framework, has
the .pyc (assuming this
tfiala added a comment.
In http://reviews.llvm.org/D19215#403844, @labath wrote:
> So, I think I have already fixed the problem with r266192.
I was able to replicate just prior to applying the change, but I'll double
check.
Regarding r266192 (which came through when I wasn't looking, sorry),
labath added a comment.
So, I think I have already fixed the problem with r266192.
This seems a bit like using a sledgehammer to kill a fly. If anything, I'd add
an assert somewhere that makes sure the file names are in the right form to
begin with.
http://reviews.llvm.org/D19215
_
tfiala created this revision.
tfiala added a reviewer: labath.
tfiala added a subscriber: lldb-commits.
Ensure all uses of a test filename in the test event infrastructure normalize
the test filename to end in the ".py" extension. Something here changed
recently such that normal code paths thro
18 matches
Mail list logo
