(And side note: if you're pushing a "lambda: self.foo()" with no arguments,
the lambda is unneeded and you can just push "self.foo" --- that cleanup
hook pushed on most tests at the end of the file is a perfect example of an
unneeded level of lambda indirection).
On Wed, Oct 21, 2015 at 12:04 PM,
On Thu, Oct 22, 2015 at 11:15 AM, Zachary Turner wrote:
> For the disassembler patch, the problem is you've defaulted the destructor
> but you've got a unique_ptr to a forward declared class. MSVC is actually
> correct in failing to compile this, and I'm not sure why other
Hi, Zachary!
On Thu, Oct 22, 2015 at 9:34 AM, Zachary Turner wrote:
> I'm taking a look now. In the future please revert changes that break a
> buildbot. If you think you might know what the fix is (i.e. you guess that
> including a header will fix it), then you can just
This is going in right now. As it is a fairly large change, it wouldn't
surprise me if someone encounters an issue. I tested this everywhere I can
and it seems fine, so please let me know if anyone encounters anything.
On Thu, Oct 22, 2015 at 11:39 AM Todd Fiala wrote:
>
The classes appear to be all forward declared intentionally, so you'd need
to include a lot of missing files. Just moving the destructor to cpp file
should be pretty easy and still allow to keep the header footprint down,
which speeds up build times.
On Thu, Oct 22, 2015 at 11:22 AM Eugene
We could also then remove unittest2 from inclusion in the lldb repo.
On Thu, Oct 22, 2015 at 11:28 AM, Todd Fiala wrote:
> I'd be okay with that.
>
> The unittest2 stuff looks like it was a vestige of being incorporated
> before unittest2 was stock (unitest) on Python
For the disassembler patch, the problem is you've defaulted the destructor
but you've got a unique_ptr to a forward declared class. MSVC is actually
correct in failing to compile this, and I'm not sure why other compilers
are accepting it. My guess is something to do with the order of includes
Yeah I think the biggest thing I wanted to check there was that there
wasn't any unittest2 behavior present in that cut of unittest2 that didn't
make it into the revamped version brought into the python distributions
when they upgraded unittest. Then it's just a big rename exercise on
replacing
Cool! I probably won't delete it from the repo entirely, because that
entails mucking with the command line options of dotest to remove any
related options, and any initialization code in dotest.py or TestBase
subclasses related to unittest2. For now I'll just delete the imports from
each
I'm taking a look now. In the future please revert changes that break a
buildbot. If you think you might know what the fix is (i.e. you guess that
including a header will fix it), then you can just check it in as a test.
If after a few attempts you still can't figure it out, then revert the
I believe it would import lldb correctly. I don't tend to run the tests
individually, but if it did work, I would use it more.
> On Oct 22, 2015, at 9:26 AM, Zachary Turner via lldb-dev
> wrote:
>
> Todd, Greg, can you guys confirm this is true? The import lldb
Todd, Greg, can you guys confirm this is true? The import lldb would
succeed if someone had their PYTHONPATH set up just right, but if really
none of us care about it, I'm with Tamas in that I'd rather remove it.
On Thu, Oct 22, 2015 at 2:55 AM Tamas Berghammer
wrote:
I plan to put this in today. Greg, should I just go ahead and delete all
the unittest2 stuff entirely? TBH I'm all for anything that reduces the
complexity of the test suite. It's got a couple hundred options that
nobody uses, seems like we should start whittling away at stuff that
doesn't get
It looks like you're generating these diffs from an SVN repo, so I'm seeing
lldb/trunk in the paths. Because of that, I can't apply it on my git repo
which doesn't have the "trunk" folder. Can you generate the diffs for the
2 aforementioned CLs by cding into the trunk directory and generating
14 matches
Mail list logo