Trunk broken with v8? (engine_v8.cpp:129:27: error: variable 'scp' set but not used)

[Trevor Robinson]

Is the trunk build known to be broken when building with v8? I tried
last week and just now, and get the -Werror failures below.

I'm primarily interested in helping with ARM support (e.g.
SERVER-1811), so I assume trunk is what I should be building. Clearly,
I'm further assuming that v8 is the best option for ARM, but let me
know if I need to use Spider Monkey instead.

$ scons --usev8
...
g++ -o build/linux2/usev8/mongo/scripting/engine_v8.o -c -Wnon-virtual-
dtor -fPIC -fno-strict-aliasing -ggdb -pthread -Wall -Wsign-compare -
Wno-unknown-pragmas -Winvalid-pch -Werror -pipe -fno-builtin-memcmp -
O3 -D_SCONS -DMONGO_EXPOSE_MACROS -D_FILE_OFFSET_BITS=64 -Ibuild/
linux2/usev8 -Isrc -Ibuild/linux2/usev8/mongo -Isrc/mongo -Ibuild/
linux2/usev8/third_party/pcre-7.4 -Isrc/third_party/pcre-7.4 -I/root/
v8/include src/mongo/scripting/engine_v8.cpp
src/mongo/scripting/engine_v8.cpp: In function 'v8::Handle<v8::Value>
mongo::namedSet(v8::Local<v8::String>, v8::Local<v8::Value>, const
v8::AccessorInfo&)':
src/mongo/scripting/engine_v8.cpp:129:27: error: variable 'scp' set
but not used [-Werror=unused-but-set-variable]
src/mongo/scripting/engine_v8.cpp: In function
'v8::Handle<v8::Boolean> mongo::namedDelete(v8::Local<v8::String>,
const v8::AccessorInfo&)':
src/mongo/scripting/engine_v8.cpp:151:27: error: variable 'scp' set
but not used [-Werror=unused-but-set-variable]
src/mongo/scripting/engine_v8.cpp: In function
'v8::Handle<v8::Boolean> mongo::indexedDelete(uint32_t, const
v8::AccessorInfo&)':
src/mongo/scripting/engine_v8.cpp:195:27: error: variable 'scp' set
but not used [-Werror=unused-but-set-variable]
src/mongo/scripting/engine_v8.cpp: In function 'v8::Handle<v8::Value>
mongo::indexedSet(uint32_t, v8::Local<v8::Value>, const
v8::AccessorInfo&)':
src/mongo/scripting/engine_v8.cpp:224:27: error: variable 'scp' set
but not used [-Werror=unused-but-set-variable]
cc1plus: all warnings being treated as errors

scons: *** [build/linux2/usev8/mongo/scripting/engine_v8.o] Error 1
scons: building terminated because of errors.

Thanks,
Trevor

[Eliot Horowitz]

Its working in buildbot,
http://buildbot.mongodb.org/waterfall

I think we're using just about trunk of v8 to fix some memory issues.
That's probably the issue.

There are still issues with v8, so definitely be careful when using it.

> --
> You received this message because you are subscribed to the Google Groups "mongodb-dev" group.
> To post to this group, send email to mongo...@googlegroups.com.
> To unsubscribe from this group, send email to mongodb-dev...@googlegroups.com.
> For more options, visit this group at http://groups.google.com/group/mongodb-dev?hl=en.
>

[Trevor Robinson]

Thanks for the link. If I'm looking at the right machine, it seems to be having out of disk space issues, though it did compile successfully a few days ago:

http://buildbot.mongodb.org/builders/Nightly%20Linux%2064-bit%20v8 

Is there any way to find out what version of GCC it's using? I didn't see that in the stdio. I figure that is most likely the issue. I'm using 4.6.1. I notice the build slave is using Python 2.5.1, while I'm using 2.7.2, so it's probably an older Linux distro.

Do you have any policies regarding GCC versions (e.g. what you build on, what you accept bugs/patches for, etc.)?

Thanks,

Trevor

[Eliot Horowitz]

We have machines building on 4.2+
4.6 should be fine

The v8 machine is on gcc 4.4.5 right now

[Trevor Robinson]

I installed gcc 4.5.4 and it worked. I've heard from others that 4.6 has new and stricter warnings.

Should I submit a bug about the 4.6 incompatibility? While using 4.5 might be a reasonable workaround for x86, there are many important ARM-specific enhancements in 4.6+ that will likely not be backported to 4.5.

[Dwight Merriman]

sure.