SURVEY:SUMMARY:BUILD_DIFFICULTY[not_applicable, reasonable_effort, code_problematic or string] reasonable_effort
SURVEY:SUMMARY:CLASSIFICATION[practical,theoretical,hardware] practical
SURVEY:SUMMARY:CORRECT_CODE_LOCATION[string] It's unclear based on the data. Just says that the git repo was found via the Stanford RAMCloud site. The data doesn't record the commit that was used for build. It seems likely that the current master branch at the time of the study was used.
SURVEY:SUMMARY:PUBLISHED_CODE[not_applicable, yes, no] yes
SURVEY:SUMMARY:SAME_VERSION[not_applicable, yes, no_but_available, no_and_not_available] yes
SURVEY:SUMMARY:STUDY_FOUND_CORRECT_CODE[not_applicable, yes, no] yes
SURVEY:AUTHOR1:BUILD_COMMENT[string] 
SURVEY:AUTHOR1:BUILD_DIFFICULTY[not_applicable, reasonable_effort, code_problematic or string] not_applicable
SURVEY:AUTHOR1:BUILD_DIFFICULTY_COMMENT[string] none
SURVEY:AUTHOR1:CLASSIFICATION[practical,theoretical,hardware] practical
SURVEY:AUTHOR1:CLASSIFICATION_COMMENT[string] 
SURVEY:AUTHOR1:CORRECT_CODE_LOCATION[string] It's unclear based on the data. Just says that the git repo was found via the Stanford RAMCloud site. The data doesn't record the commit that was used for build. It seems likely that the current master branch at the time of the study was used.
SURVEY:AUTHOR1:PUBLIC_COMMENT[string] 
SURVEY:AUTHOR1:PUBLISHED_CODE[not_applicable, yes, no] yes
SURVEY:AUTHOR1:SAME_VERSION[not_applicable, yes, no_but_available, no_and_not_available] yes
SURVEY:AUTHOR1:SAME_VERSION_COMMENT[string] The SOSP11 paper's code is available through the sosp2011-camera-ready tag in our public git repo.
SURVEY:AUTHOR1:STUDY_FOUND_CORRECT_CODE[not_applicable, yes, no] no
SURVEY:AUTHOR2:BUILD_COMMENT[string] It looks like there were a couple of problems related to external packages used by RAMCloud. In one case it appears that Protocol Buffers had not been installed. In another case there appear to be problems with gtest. I don't know why the gtest problems are happening, but I do know that many people outside Stanford have successfully compiled and run RAMCloud. There is even a mailing list where you can ask for help.
SURVEY:AUTHOR2:BUILD_DIFFICULTY[not_applicable, reasonable_effort, code_problematic or string] reasonable_effort
SURVEY:AUTHOR2:BUILD_DIFFICULTY_COMMENT[string] 
SURVEY:AUTHOR2:CLASSIFICATION[practical,theoretical,hardware] practical
SURVEY:AUTHOR2:CLASSIFICATION_COMMENT[string] 
SURVEY:AUTHOR2:CORRECT_CODE_LOCATION[string] 
SURVEY:AUTHOR2:PUBLIC_COMMENT[string] 
SURVEY:AUTHOR2:PUBLISHED_CODE[not_applicable, yes, no] yes
SURVEY:AUTHOR2:SAME_VERSION[not_applicable, yes, no_but_available, no_and_not_available] yes
SURVEY:AUTHOR2:SAME_VERSION_COMMENT[string] 
SURVEY:AUTHOR2:STUDY_FOUND_CORRECT_CODE[not_applicable, yes, no] yes
SURVEY:AUTHOR3:BUILD_COMMENT[string] I think the build error would be obvious to anyone that's familiar with building C/C++ programs, and I don't think the conclusion of a build failure for this is legitimate.<br/><br/>From the build notes:<br/>http://reproducibility.cs.arizona.edu/v2/data/sosp11_OngaroRSOR11_build.txt<br/>The first error a few lines in is:<br/>./Core/ProtoBuf.h:16:37: fatal error: google/protobuf/message.h: No such file or directory<br/>compilation terminated.<br/><br/>So Google's protocol buffers library wasn't found: either it wasn't installed or wasn't at the expected location. <br/><br/>See also how Peter Bailis was able to build the project: https://github.com/shriram/repro-in-cs/blob/master/data/sosp11/OngaroRSOR11/cleared.txt
SURVEY:AUTHOR3:BUILD_DIFFICULTY[not_applicable, reasonable_effort, code_problematic or string] reasonable_effort
SURVEY:AUTHOR3:BUILD_DIFFICULTY_COMMENT[string] 
SURVEY:AUTHOR3:CLASSIFICATION[practical,theoretical,hardware] practical
SURVEY:AUTHOR3:CLASSIFICATION_COMMENT[string] 
SURVEY:AUTHOR3:CORRECT_CODE_LOCATION[string] 
SURVEY:AUTHOR3:PUBLIC_COMMENT[string] RAMCloud is a large system, and we've made a huge effort to produce some quality open-source code. Several factors fundamentally make it harder to reproduce RAMCloud's fast crash recovery results: RAMCloud had to be in C++ for the level of performance we were looking for, making it less portable and harder to build than most modern languages, and it needs a large cluster with very modern hardware. We expect our first users to have to configure a bunch of hardware to use RAMCloud effectively, so making it build without effort on a variety of distros wasn't high on our list. That's a place where leveraging a growing community makes sense: it's really easy for outsiders to contribute updated build instructions and patches needed to build RAMCloud on their systems, whereas we only develop with one particular cluster configuration. RAMCloud is an ongoing project, and we hope building and running it gets easier over time. For example, Jakob Blomer has recently developed some RPM packages in hopes of making it easier for others to get started.
SURVEY:AUTHOR3:PUBLISHED_CODE[not_applicable, yes, no] yes
SURVEY:AUTHOR3:SAME_VERSION[not_applicable, yes, no_but_available, no_and_not_available] yes
SURVEY:AUTHOR3:SAME_VERSION_COMMENT[string] It's an ever-progressing git repo, but that version is in the history somewhere.
SURVEY:AUTHOR3:STUDY_FOUND_CORRECT_CODE[not_applicable, yes, no] yes