zturner (Zachary Turner)
User

Projects

User does not belong to any projects.

User Details

User Since
May 26 2014, 12:49 PM (182 w, 3 d)

Recent Activity

Yesterday

zturner added a comment to D40366: [Support][Parallel] ThreadPoolExecutor fixes for Windows.

I absolutely do not want to disable support for ConcRT under any circumstances unless we find that it is a completely broken library, which I highly doubt. Even then, I would want to ask Microsoft to fix it.

What is the exact problem you're running into?

Hi, sorry, I thought that this would be a "private" patch. Don't worry, I do not intend to make any changes without proper review!

Ahh, sorry about that. One of the Phabricator triggers must have added me automatically.

Wed, Nov 22, 11:18 AM
zturner requested changes to D40366: [Support][Parallel] ThreadPoolExecutor fixes for Windows.

I absolutely do not want to disable support for ConcRT under any circumstances unless we find that it is a completely broken library, which I highly doubt. Even then, I would want to ask Microsoft to fix it.

Wed, Nov 22, 10:46 AM

Tue, Nov 21

zturner added a comment to D40328: Use a single file name to represent a lock.

Out of curiosity, what do we use LockFileManager for in LLVM, and what are you trying to use it for in this specific use case? Because of certain quirks in Windows file system semantics, lock files are in general not a robust solution most of the time -- maybe even ever.

Tue, Nov 21, 6:49 PM
zturner added reviewers for D40328: Use a single file name to represent a lock: pcc, amccarth.
Tue, Nov 21, 6:46 PM

Mon, Nov 20

zturner committed rL318722: Re-revert "Refactor debuginfo-tests.".
Re-revert "Refactor debuginfo-tests."
Mon, Nov 20, 5:21 PM
zturner committed rL318697: Resubmit "Refactor debuginfo-tests" again..
Resubmit "Refactor debuginfo-tests" again.
Mon, Nov 20, 1:43 PM

Fri, Nov 17

zturner committed rL318564: Fix an issue with llvm lit tool substitutions..
Fix an issue with llvm lit tool substitutions.
Fri, Nov 17, 2:52 PM
zturner accepted D40142: [cmake] Use llvm-lit directory when provided for stand-alone build.
Fri, Nov 17, 2:13 PM
zturner created D40192: WritableMemoryBuffer.
Fri, Nov 17, 1:07 PM
zturner accepted D40188: Enable PDB generation with lld in asan and cfi tests on Windows..
Fri, Nov 17, 11:36 AM
zturner accepted D39567: [lit] Implement non-pipelined ‘mkdir’, ‘diff’ and ‘rm’ commands internally.

looks good to me, rnk@ are you ok?

Fri, Nov 17, 9:50 AM
zturner added a comment to D40170: [llvm-tblgen] - Stop using std::string in RecordKeeper..

There's been some talk about making LLVM_OPTIMIZED_TABLEGEN the default, but it hasn't happened for whatever reason.

Fri, Nov 17, 9:49 AM

Thu, Nov 16

zturner committed rL318489: Don't #include MemoryBuffer.h from Host.h..
Don't #include MemoryBuffer.h from Host.h.
Thu, Nov 16, 5:00 PM
zturner committed rL318486: Re-revert "Refactor debuginfo-tests".
Re-revert "Refactor debuginfo-tests"
Thu, Nov 16, 4:41 PM
zturner committed rL318462: Fix some undefined beahvior in FileMapping..
Fix some undefined beahvior in FileMapping.
Thu, Nov 16, 2:40 PM
zturner committed rL318435: Resubmit "Refactor debuginfo-tests".
Resubmit "Refactor debuginfo-tests"
Thu, Nov 16, 10:26 AM

Wed, Nov 15

zturner added a comment to D39994: Loosen MSVC 2017 path requirements.

I'm not suuuper opposed, but at the same time if this code is bothering people (and it is, consistently), I don't changing the requirements from "confusing rule A" to "confusing rule B" is going to solve the long term burden that people keep running into.

Wed, Nov 15, 6:34 PM
zturner added a comment to D40079: Make sure DataBufferLLVM contents are writable.

On second thought I actually think the strategy used here is fine. But we have llvm::WritableBinaryStream which could server as a single base interface for a mapped_file_region based implementation as well as a malloc-based implementation.

Wed, Nov 15, 4:36 PM
zturner accepted D39517: [DebugInfo/PDB] Adding getUndecoratedNameEx and IPDB interfaces for IDiaEnumTables and IDiaTable..

We'll take a look at what we can do with llvm-pdbutil to test these changes. Right now we are only using publicly available information and I'm not sure I can say much more about what DIA SDK does beyond what you can already see online. I'll ask if there is some better information we can share.

Wed, Nov 15, 9:41 AM

Mon, Nov 13

zturner committed rL318112: Revert "Update test_debuginfo.pl script to point to new tree location.".
Revert "Update test_debuginfo.pl script to point to new tree location."
Mon, Nov 13, 3:34 PM

Fri, Nov 10

zturner committed rL317949: Update test_debuginfo.pl script to point to new tree location..
Update test_debuginfo.pl script to point to new tree location.
Fri, Nov 10, 3:13 PM
zturner added a comment to D39896: Remove last Host usage from ArchSpec.

Drive by comment:

You're right, the Triple stuff is in ADT! I was using it as an example of something you clearly wouldn't do so I didn't actually check to see if it was done. That's pretty funny.

I thought that ADT stood for Abstract Data Types - though I actually I don't remember where I got that impression... Having Triple there does seem confusing. That is a "how we specify Targets type thing", not a fancy data type type thing... I would never think to look for that facility alongside StringRef & DenseMap etc.

Triple.h is in ADT; Triple.cpp is in Support. I agree it belongs in Support, and I'm not sure why it's schizo like that.

Fri, Nov 10, 2:25 PM
zturner committed rL317931: Fix for skipped CMake configuration on debuginfo-tests..
Fix for skipped CMake configuration on debuginfo-tests.
Fri, Nov 10, 2:12 PM
zturner committed rL317925: [debuginfo-tests] Make debuginfo-tests work in a standard configuration..
[debuginfo-tests] Make debuginfo-tests work in a standard configuration.
Fri, Nov 10, 12:58 PM
zturner closed D39605: [debuginfo-tests] Decouple debuginfo-tests from the clang source tree. by committing rL317925: [debuginfo-tests] Make debuginfo-tests work in a standard configuration..
Fri, Nov 10, 12:58 PM
zturner added a comment to D39885: Disable GC and ICF when /debug is present.

Seems like a good idea to me, for consistency.

Fri, Nov 10, 11:28 AM
zturner added a comment to D39896: Remove last Host usage from ArchSpec.

I'd be open to having another organizational component that isn't Utility. But as Jim said, there isn't a critical mass of stuff yet in Utility to figure out where it makes sense to draw the line.

Fri, Nov 10, 11:26 AM

Thu, Nov 9

zturner accepted D39876: [support] allocate exact size required for mapping in Support/Windws/Path.inc.

yay, huge space gains ftw.

Thu, Nov 9, 3:58 PM
zturner added a comment to D39737: [MC] Fix regression tests on Windows when git “core.autocrlf” is set to true.

Ahh sorry, I may have misunderstood your previous question. I thought we were talking about setting core.autocrlf=false at the global LLVM .gitattributes file, and then not having anything special for this one particular test.

Thu, Nov 9, 3:54 PM
zturner committed rL317830: Add a Cross-compilation toolchain file for MSVC..
Add a Cross-compilation toolchain file for MSVC.
Thu, Nov 9, 12:38 PM
zturner closed D39814: Create a Cross-Compilation CMake toolchain file for NonWindows -> Windows by committing rL317830: Add a Cross-compilation toolchain file for MSVC..
Thu, Nov 9, 12:38 PM
zturner closed D39761: [Support] Make llvm::Error faster.
Thu, Nov 9, 11:32 AM
zturner committed rL317824: [Support] Make llvm::Error and Expected faster..
[Support] Make llvm::Error and Expected faster.
Thu, Nov 9, 11:32 AM
zturner updated the diff for D39814: Create a Cross-Compilation CMake toolchain file for NonWindows -> Windows.

Updated with the suggestions. The -imsvc argument is nice, and allows us to get rid of manually specifying the resource directory, which was one of the FIXMEs I had before. So that is now fixed.

Thu, Nov 9, 10:55 AM
zturner added inline comments to D39844: CompilerType: Add ability to retrieve an integral template argument.
Thu, Nov 9, 10:24 AM
zturner added a comment to D39605: [debuginfo-tests] Decouple debuginfo-tests from the clang source tree..

I'm fine with the changes being made. We just need to coordinate a date & time for the cutover so we don't set Green Dragon ablaze.

Also, someone needs to look at the buildbot configs. It has been so long since I have looked at those configs I'm not sure which bots on that side will need to be updated (if any).

Thu, Nov 9, 9:48 AM

Wed, Nov 8

zturner created D39814: Create a Cross-Compilation CMake toolchain file for NonWindows -> Windows.
Wed, Nov 8, 12:41 PM
zturner added a comment to D39809: MSVC toolchain.

Here is the output of CMake:

Wed, Nov 8, 10:44 AM
zturner created D39809: MSVC toolchain.
Wed, Nov 8, 10:41 AM
zturner accepted D38891: Implement flock for Windows in compiler-rt.
Wed, Nov 8, 9:18 AM

Tue, Nov 7

zturner updated the diff for D39761: [Support] Make llvm::Error faster.

I went ahead and applied the same optimization for llvm::Expected, and also incorporated other suggestions. Even though Expected wasn't showing up on the profile, this only arose because we happened to be a heavy user of Error. If someone else happened to be a heavy user of Expected, we'd see the same issues arise.

Tue, Nov 7, 4:50 PM
zturner created D39761: [Support] Make llvm::Error faster.
Tue, Nov 7, 1:54 PM

Mon, Nov 6

zturner accepted D39689: Add a dependency from check-lldb on lld.
Mon, Nov 6, 10:22 AM

Fri, Nov 3

zturner committed rL317380: Fix variable name..
Fix variable name.
Fri, Nov 3, 3:34 PM
zturner added a comment to D39605: [debuginfo-tests] Decouple debuginfo-tests from the clang source tree..

Definitely. I'm in no rush. I'll push out a message to llvm-dev before the day's over.

Fri, Nov 3, 2:57 PM
zturner added a comment to D39605: [debuginfo-tests] Decouple debuginfo-tests from the clang source tree..

At Sony we build our own unified tree by pasting together several subtrees, including debuginfo-tests, and moving one will cause a bit of disruption. I remember the idea of moving debuginfo-tests has come up previously, and in principle I have no problem with it, but a more explicit announcement on llvm-dev/cfe-dev would be appropriate.

Fri, Nov 3, 2:35 PM
zturner created D39605: [debuginfo-tests] Decouple debuginfo-tests from the clang source tree..
Fri, Nov 3, 10:20 AM

Thu, Nov 2

zturner added a comment to D39567: [lit] Implement non-pipelined ‘mkdir’, ‘diff’ and ‘rm’ commands internally.

Update: I actually do see where you're testing that they work, but I'm not convinced that the tests are sufficient. For several reasons.

Thu, Nov 2, 2:25 PM
zturner added a comment to D39567: [lit] Implement non-pipelined ‘mkdir’, ‘diff’ and ‘rm’ commands internally.

First off, let me say this is very cool. I look forward to the day when all commands are builtin, and we can switch every client to the builtin lit shell.

Thu, Nov 2, 2:20 PM
zturner added a comment to D39517: [DebugInfo/PDB] Adding getUndecoratedNameEx and IPDB interfaces for IDiaEnumTables and IDiaTable..

Looks good, mostly just nitpicks. Can you also add a test for this? We've checked in an several actual PDBs into test/tools/llvm-pdbutil/Inputs, but you'd need to add a command line option to the pretty mode of llvm-pdbutil so that it can dump it. Perhaps making it a hidden option. Hopefully the PDBs that are checked in already will cause DIA to return something interesting for the tables.

Thu, Nov 2, 1:33 PM
zturner added inline comments to D38829: Python: SetOutputFileHandle doesn't work with IOBase.
Thu, Nov 2, 1:17 PM
zturner added a comment to D38829: Python: SetOutputFileHandle doesn't work with IOBase.

Ok, I wasn't aware of the libedit problem.

Thu, Nov 2, 11:52 AM
zturner added inline comments to D38829: Python: SetOutputFileHandle doesn't work with IOBase.
Thu, Nov 2, 11:16 AM
zturner added inline comments to D38829: Python: SetOutputFileHandle doesn't work with IOBase.
Thu, Nov 2, 11:15 AM
zturner requested changes to D38829: Python: SetOutputFileHandle doesn't work with IOBase.
Thu, Nov 2, 10:42 AM
zturner added inline comments to D38829: Python: SetOutputFileHandle doesn't work with IOBase.
Thu, Nov 2, 10:41 AM
zturner added a comment to D39539: [llvm-nm] Don't error out on multiple occurrances of the -g/--external-only flag.

Agreed. Optional should basically be called OneOrZero. And if you think about it in those terms, it obviously doesn't make sense to allow it to be specified more than once.

Thu, Nov 2, 9:15 AM

Wed, Nov 1

zturner added inline comments to D39436: Add regex support to file (-f) and module (-s) breakpoint options..
Wed, Nov 1, 6:15 PM
zturner added inline comments to D39436: Add regex support to file (-f) and module (-s) breakpoint options..
Wed, Nov 1, 5:30 PM
zturner accepted D39465: Add feature to determine if host architecture is 64-bit in llvm-lit.

I actually think that the llvm-64-bits feature should go in the common config.py setup, but I don't think we're quite there yet. It would require some additional work to make sure everyone who uses LLVM can agree on a common set of variables that must be passed in through configure, and currently we don't have that.

Wed, Nov 1, 5:24 PM
zturner updated subscribers of D39517: [DebugInfo/PDB] Adding getUndecoratedNameEx and IPDB interfaces for IDiaEnumTables and IDiaTable..

Nice! I'm looking forward to reviewing this tomorrow.

Wed, Nov 1, 4:06 PM
zturner added a comment to D39436: Add regex support to file (-f) and module (-s) breakpoint options..

I was unhappy when we went over two pointers for a FileSpec when m_syntax was added due to the extra size. Anything we can do to make this smaller would be great, so the type on the enum would work, but as you say the alignment will nullify that. The two ConstString members contain a pointer which isn't aligned so we can't use any bits from the low end of the pointer. Are there any classes that take advantage of high bits in pointers? Most if not all OS's don't use the entire 64 bit address space... It would be great to get lldb_private::FileSpec down to just 2 pointers again.

ConstString doesn't *currently* contain aligned pointers, but there's no reason we couldn't make it contain aligned pointers. Then we could use llvm::PointerUnion.

I would be fine with that.

That said, I want to state again that I think this change is the wrong direction. I don't think we need this functionality in FileSpec, or even in another class. I think it is better served in the script.

Agreed as well. I do like the idea of source path regexes, but only if we have a real need to add it to the API.

I haven't had time to really look into this, but it seems that maintaining two independent strings, one for directory and one for basename, is just for convenience. We could easily keep it in a single string with an index to basename. When the directory portion is updated, e.g., resolved, etc., we just overwrite the string and adjust the index. And adjust accessors as needed.

Wed, Nov 1, 11:09 AM
zturner added a comment to D39436: Add regex support to file (-f) and module (-s) breakpoint options..

I was unhappy when we went over two pointers for a FileSpec when m_syntax was added due to the extra size. Anything we can do to make this smaller would be great, so the type on the enum would work, but as you say the alignment will nullify that. The two ConstString members contain a pointer which isn't aligned so we can't use any bits from the low end of the pointer. Are there any classes that take advantage of high bits in pointers? Most if not all OS's don't use the entire 64 bit address space... It would be great to get lldb_private::FileSpec down to just 2 pointers again.

Wed, Nov 1, 9:52 AM

Tue, Oct 31

zturner accepted D39482: Add system-linux to allow tests run with llvm-lit to restrict themselves to linux.
Tue, Oct 31, 4:10 PM
zturner added inline comments to D39465: Add feature to determine if host architecture is 64-bit in llvm-lit.
Tue, Oct 31, 11:41 AM
zturner accepted D39449: Rewrite FileOutputBuffer as two separate classes..
Tue, Oct 31, 11:28 AM
zturner added inline comments to D39449: Rewrite FileOutputBuffer as two separate classes..
Tue, Oct 31, 11:12 AM

Mon, Oct 30

zturner added inline comments to D39449: Rewrite FileOutputBuffer as two separate classes..
Mon, Oct 30, 8:21 PM
zturner added inline comments to D39444: [Support] Make the default chunk size of raw_fd_ostream to 1 GiB..
Mon, Oct 30, 7:37 PM
zturner added a comment to D39263: [support] remove tautological comparison in Support/Windows/Path.inc.

@zturner do you want me to make the semantic changes in the same commit or can we ship the size check removal first and then change the functionality? The latter would be my preference.

Mon, Oct 30, 7:33 PM
zturner added inline comments to D39444: [Support] Make the default chunk size of raw_fd_ostream to 1 GiB..
Mon, Oct 30, 7:32 PM
zturner added a comment to D39436: Add regex support to file (-f) and module (-s) breakpoint options..

Zachary's suggestion is better than adding regex patterns to FileSpec, but I still don't like the idea of encoding option types in the option values.

Are you talking about fnmatch? Is that portable? If not, i can rewrite it,.

Moreover, this doesn't really apply to the main use of -f - file & line breakpoints. The only plausible use for a source file filter for file & line breakpoints is when you want to restrict breaking on inlined code to a subset of the files that inline it. But in that case you would want a separate option, since you need to specify the inlined file = with the -f option - and the inlining files - with some other option.

Wouldn't it be better to add a --source-file-regex option, pick some free single letter, and use that if you want to provide a pattern for the "-p" breakpoint option. Then we could also use it when specifying some other file filter.

This is valid for all breakpoints, not just those created via the '-p' option. I can gen up patch that add --source-file-regex to all breakpoints, but I also want to use it from python, so that api change might be problematic -- can we add additional default parameters without breaking ABI?

Btw, here's an example of how I want to use it -- cuts time to create breakpoint by over half, from 15 seconds down to 7 -- which makes my clangdiag module load much faster:

br s -n "DiagnosticsEngine::Report" -f "regex:.*/clang/.*"
Mon, Oct 30, 5:26 PM
zturner added a comment to D39436: Add regex support to file (-f) and module (-s) breakpoint options..

Zachary's suggestion is better than adding regex patterns to FileSpec, but I still don't like the idea of encoding option types in the option values.

Moreover, this doesn't really apply to the main use of -f - file & line breakpoints. The only plausible use for a source file filter for file & line breakpoints is when you want to restrict breaking on inlined code to a subset of the files that inline it. But in that case you would want a separate option, since you need to specify the inlined file = with the -f option - and the inlining files - with some other option.

Wouldn't it be better to add a --source-file-regex option, pick some free single letter, and use that if you want to provide a pattern for the "-p" breakpoint option. Then we could also use it when specifying some other file filter.

Mon, Oct 30, 4:45 PM
zturner added a comment to D39436: Add regex support to file (-f) and module (-s) breakpoint options..

Hmm, weird. Maybe it already is? Even though I didn't see it on the original email. Anyway, ignore my last suggestion.

On to actual comments: I'm not really crazy about the regex: prefix. Seems ugly and non-intuitive. Couldn't we just treat every filename as an argument to fnmatch (3)? Then you don't need any weird prefixes. Besides, regex:foo is a valid filename so if we're going to be making it so that some arguments to -f all of a sudden have to be escaped because they're no longer interpreted literally, then we might as well do it in a way that's intuitive and people are already familiar with.

The problem with -f and -s is that they create FileSpec's, and FileSpec has some special handling for paths, so I'm not sure how to handle this without a prefix. Is there a delimiter that's invalid on all platforms? I thought ':' would work since we use it for path delimiters, but I've only got a Mac to test on right now. I suppose we could get really ugly and use a uri style prefix, e.g., "regex://".

Mon, Oct 30, 4:20 PM
zturner added a comment to D39436: Add regex support to file (-f) and module (-s) breakpoint options..

Hmm, weird. Maybe it already is? Even though I didn't see it on the original email. Anyway, ignore my last suggestion.

Mon, Oct 30, 4:01 PM
zturner added a comment to D39436: Add regex support to file (-f) and module (-s) breakpoint options..

Would you mind deleting and re-creating this revision with lldb-commits added as a subscriber? I don't think it's sufficient to just "add" it as a subscriber after the fact, I think it has to be done as part of the initial creation, or for some reason it won't show up on the mailing list.

Mon, Oct 30, 3:58 PM

Fri, Oct 27

zturner accepted D39387: Invert ArchSpec<->Platform dependency.
Fri, Oct 27, 4:01 PM
zturner added inline comments to D39387: Invert ArchSpec<->Platform dependency.
Fri, Oct 27, 3:48 PM
zturner added inline comments to D39387: Invert ArchSpec<->Platform dependency.
Fri, Oct 27, 2:27 PM
zturner added inline comments to D39387: Invert ArchSpec<->Platform dependency.
Fri, Oct 27, 2:25 PM
zturner committed rL316798: Force #define GTEST_LANG_CXX11..
Force #define GTEST_LANG_CXX11.
Fri, Oct 27, 2:13 PM
zturner accepted D39263: [support] remove tautological comparison in Support/Windows/Path.inc.
Fri, Oct 27, 2:08 PM
zturner added inline comments to D39387: Invert ArchSpec<->Platform dependency.
Fri, Oct 27, 2:06 PM

Thu, Oct 26

zturner added a comment to D39315: Correct the start address for exported windows functions in arm.

So when you're using ds2 as the remote, are you still using the normal lldb test suite? dotest.py? If so, then we could have a test decorator that says @expectedFailure(not(debugserver=ds2)) or something. Then, even though you're the only one that can run it, at least YOU are sure it works. Some coverage is better than nothing. Is something like this possible?

Thu, Oct 26, 11:32 AM
zturner added a comment to D39315: Correct the start address for exported windows functions in arm.
In D39315#908246, @sas wrote:

Same thing here, I have no idea how to go about testing something specific like this. Given that this is Windows Phone, it's even worse than simply Windows Desktop because the only way to test is by doing remote debugging. I suppose checking in binaries to the tree so we can analyze the symbols and ensure correctness is the only way to do it?

Thu, Oct 26, 11:12 AM
zturner added a comment to D39314: Implement basic DidAttach and DidLaunch for DynamicLoaderWindowsDYLD.

Ahh, you might also run the exact same test again but where you've loaded this inside of main with LoadLibrary instead of relying on early binding by the loader.

Thu, Oct 26, 10:56 AM
zturner added a comment to D39314: Implement basic DidAttach and DidLaunch for DynamicLoaderWindowsDYLD.

A test would be something like this:

Thu, Oct 26, 10:55 AM
zturner added a comment to D36347: New lldb python module for adding diagnostic breakpoints.

Do I understand correctly that this will insert breakpoints on *all* clang diagnostics? That's not necessarily bad, but I was under the impression originally that it would let you pick the diagnostics you wanted to insert breakpoints on.

Thu, Oct 26, 10:45 AM

Wed, Oct 25

zturner accepted D39278: [lit] Respect LLVM_LIT_TOOLS_DIR when looking for 'tar' on Windows..

We really should be using llvm_config.get_process_output() here, fwiw, but as you said, this patch is just fixing an execution failure.

Wed, Oct 25, 7:12 AM · lld

Oct 24 2017

zturner added a comment to D39263: [support] remove tautological comparison in Support/Windows/Path.inc.

Let me know if this is correct. I think we can actually get (potentially very large) savings from this.

Oct 24 2017, 5:07 PM
zturner added reviewers for D39263: [support] remove tautological comparison in Support/Windows/Path.inc: rnk, majnemer, compnerd.

Adding more Windows people, because as I look over this code, it actually looks wrong to me. The value we pass for MaximumFileSize seems unnecessarily large. I'm going through the docs at the moment to figure out the correct thing to do here.

Oct 24 2017, 5:00 PM
zturner accepted D31172: Move stop info override callback code from ArchSpec into Process.

I will ping them for some numbers and more details of their test setup. Regardless, I didn't mean to derail the code review. But, I really really want to reach a point where we can stop falling back on the "we need to be safe even in the presence of stuff that is clearly not user input" argument. I understand the concerns, but I don't think this is a reasonable path forward for the project. If it's not user input, if we own it, then we can make whatever assumption we want that leads to the best performance and memory usage characteristics.

Oct 24 2017, 3:44 PM
zturner added inline comments to D31172: Move stop info override callback code from ArchSpec into Process.
Oct 24 2017, 3:17 PM
zturner added inline comments to D31172: Move stop info override callback code from ArchSpec into Process.
Oct 24 2017, 2:48 PM
zturner added inline comments to D31172: Move stop info override callback code from ArchSpec into Process.
Oct 24 2017, 2:47 PM
zturner added inline comments to D31172: Move stop info override callback code from ArchSpec into Process.
Oct 24 2017, 2:34 PM
zturner added a comment to D31172: Move stop info override callback code from ArchSpec into Process.

I know you're doing things the way it's always been done, but I want to start questioning some long-standing practices :) So I'm not picking on you specifically, but anywhere we can migrate towards something better incrementally, I think we should do so.

Oct 24 2017, 2:08 PM
zturner added a comment to D39215: Default to using in-tree clang for building test executables.
In D39215#905259, @ted wrote:

We build lldb, clang and tools for Hexagon only, and call them hexagon-lldb, hexagon-clang, etc. The test infrastructure is smart enough to pick up hexagon-lldb-mi if we tell it to run with hexagon-lldb using --executable; will it be smart enough to run an in-tree hexagon-clang?

@labath, we run on Windows using hexagon-clang and hexagon-clang++; don't forget the embedded cases when choosing compilers and running tests.

I'm all for removing redundant variables.

Oct 24 2017, 9:02 AM

Oct 23 2017

zturner added a comment to D39215: Default to using in-tree clang for building test executables.

Ok the issue is that you cant use CMake generator expressions in this way. This should work though:

Oct 23 2017, 7:20 PM
zturner accepted D39162: [test] Fix clang-test for FreeBSD and NetBSD.
Oct 23 2017, 6:17 PM