- User Since
- Mar 23 2016, 8:38 AM (273 w, 6 d)
@kastiglione Good point, I dropped using any shell at the source.
- Addressing Adrian's comments (sorry for the delay)
@vsk Just fixing this to get the matrix bot green when it gets back online, but you might have a better idea how to fix this (like, making the call site info decorator more strict?). FWIW, all the failing tests with their output can be found here: https://ci.teemperor.de/job/llvm-master-test-matrix/30/
FWIW, I'm open to preferring bash over zsh as the default shell. But as zsh is already the default on macOS we might as well start using it when possible.
- Add pipe and backslash
- Upload the fish diff
Whoops, wrong diff...
- Add pipe and backslash also as special characters.
ProgressEventThreadFunction is now unused which will raise a warning (and break -Werror builds):
Sorry, feel free to me directly sooner :)
Could you move this function into the Debugger class and just make SBDebugger forward to that function? We usually keep the SB* classes as thin wrappers that only contain binding specific logic and have the actual implementation in the sibling class without the SB prefix.
- Remove $ so that env variables will be expanded.
LGTM, thanks for cleaning this up.
Thu, Jun 17
LGTM now, just a small nit about the redundant "error: " string.
There are two places where this isn't redundant (see inline comments) but otherwise this LGTM. Thanks!
- Added a unit test (thanks Shafik!)
Thanks for writing tests, it's really appreciated! FWIW, I think that you can just check in new tests without having to go through a full review (unless you do want feedback for it).
Wed, Jun 16
LGTM and passes on macOS and Linux for me. Thanks!
At least on the test suite on Linux that branch is anyway never taken, so LGTM. Thanks!
There is some cleanup that needs to be done before this can land. https://reviews.llvm.org/D104374 fixes some deprecation warnings from Python and there are some macOS APIs that got deprecated (which I still need to fix)
Tue, Jun 15
- assert -> lldbassert
In fact, the same idea also occurred to me before. I think it's also achievable by a plugin sphinx-code-tabs.
Personally I think if there is the pointer (and sizeof(std::unique_ptr<XXX, YYY> == 16) there should be the "deleter =" specified even if it is default deleter.
I think the way the provider is supposed to work is that there is always deleter child as long as it's not default_delete<T>, so I think we have to check for the name to avoid that we also hide an empty user-specified deleter.
Mon, Jun 14
Instead of copying the text from the Python page, I was actually thinking whether we should make the 'scripting' page more generic and just add the language-specific examples for Python and Lua there. I am not sure what's the best way to do that visually though with RST. We could just go for raw HTML and use a tab switcher like this:
Not sure what else we should do to properly deprecate an API. LLVM's approach would be to put [[deprecated("bla")]] the C++ APIs (which would mean you need a C++14 compiler to use the SB API) and there is also the lldb-versioning.h header (which seems to be reserved for actual API breakages).
- Remove the deprecation attribute that got inserted by accident.
- Make language more uniform
- Point out anonymous struct is a GNU extension.
Not sure what happened to my accept, but I'll LGTM this again. Thanks!
Fri, Jun 11
LGTM, thanks! (And congrats on your first patch)
Reopen because of the revert.
LGTM, big thanks for the patch & quick turnaround!
Thu, Jun 10
I think we should instead implement the Terminate function that the plugin system provides to tear down our state. I made a patch in D104093 that
moves the once_flags to the internal state struct, so with a proper Terminate implementation + my patch this should all work properly.
This seems reasonable to me, but I'll leave this open for a while in case someone that knows more about Rust mangling shows up. Otherwise I'll accept this next week.
Sure I can take a look, but I don't see the immediate problem when looking at the backtrace.
LGTM beside some minor formatting bugs that got exacerbated by the clang-tidy run. I got bored of writing 'format' everywhere so I translated every comment into a different language via Google translate. I'll already accept this as this is easy to fix.
Wed, Jun 9
This doesn't compile for me (on Linux):