Mike Percy has posted comments on this change.

Change subject: Add Google Breakpad support to Kudu
......................................................................


Patch Set 1:

(6 comments)

http://gerrit.cloudera.org:8080/#/c/5620/1//COMMIT_MSG
Commit Message:

Line 28:   because breakpad does not properly namespace its includes (however it
> I'm not sure what you mean by "namespace" here. You don't mean namespace as
Yes, directory prefix. I actually wrote a script to do preprocessing of the 
breakpad source to include a "breakpad/" prefix in their #includes and then 
dropped it because I thought it was kind of dirty. Ideally, we could convince 
them to add a prefix but it seems like a tall order since people would also 
have to change Firefox and Chromium code after doing that. But if you think 
that hack is better then I can bring it back.


http://gerrit.cloudera.org:8080/#/c/5620/1/src/kudu/integration-tests/minidump_generation-itest.cc
File src/kudu/integration-tests/minidump_generation-itest.cc:

Line 56:   ASSERT_TRUE(found) << "Unable to find minidump after " << 
timeout.ToString();
> could you restructure this test using AssertEventually(...) to be a bit eas
Nice, done.


http://gerrit.cloudera.org:8080/#/c/5620/1/src/kudu/util/CMakeLists.txt
File src/kudu/util/CMakeLists.txt:

Line 208:   breakpad_client
> does putting this stuff in util/ mean that we're ending up linking unnecess
Not sure. I will check.


http://gerrit.cloudera.org:8080/#/c/5620/1/src/kudu/util/minidump.cc
File src/kudu/util/minidump.cc:

Line 99:     const int path_len = my_strlen(path);
> I don't think it's calling into libc that's problematic, is it? I think the
This is what it says in 
https://chromium.googlesource.com/breakpad/breakpad/+/master/src/common/linux/linux_libc_support.h
 ...

  // This header provides replacements for libc functions that we need. We if
  // call the libc functions directly we risk crashing in the dynamic linker as
  // it tries to resolve uncached PLT entries.


PS1, Line 220: CreateDirsRecursively
> we sure that recursive create is what we want vs just creating the parent a
Personally I think it makes sense because they can specify an arbitrary path 
for FLAGS_minidump_path.


http://gerrit.cloudera.org:8080/#/c/5620/1/src/kudu/util/minidump.h
File src/kudu/util/minidump.h:

Line 26: Status RegisterMinidump(const char* cmd_line_path);
> seems odd to see a C string here instead of std::string
Ah, missed this from the original Impala source. Done.


-- 
To view, visit http://gerrit.cloudera.org:8080/5620
To unsubscribe, visit http://gerrit.cloudera.org:8080/settings

Gerrit-MessageType: comment
Gerrit-Change-Id: I495695cc38b75377f20b0497093a81ed5baa887f
Gerrit-PatchSet: 1
Gerrit-Project: kudu
Gerrit-Branch: master
Gerrit-Owner: Mike Percy <[email protected]>
Gerrit-Reviewer: Adar Dembo <[email protected]>
Gerrit-Reviewer: Dinesh Bhat <[email protected]>
Gerrit-Reviewer: Kudu Jenkins
Gerrit-Reviewer: Lars Volker <[email protected]>
Gerrit-Reviewer: Mike Percy <[email protected]>
Gerrit-Reviewer: Todd Lipcon <[email protected]>
Gerrit-HasComments: Yes

Reply via email to