Prevent the preprocessor from replacing std{in,out} with macro #2111
Conversation
|
build fails, but the basic question is if we want to keep the naming (and have to fix the macro) or if there is a proper name which is not taken by a macro which can work. |
judge/runpipe.cc
Outdated
| @@ -641,22 +641,22 @@ struct state_t { | |||
| tie(read_end, write_end) = make_pipe(); | |||
| logmsg(LOG_DEBUG, "setting up pipe #%ld (fd %d) -> proxy (fd %d)", i, | |||
| write_end, read_end); | |||
| process.stdout = write_end; | |||
| process.pstdout = write_end; | |||
There was a problem hiding this comment.
What does p stand for? I think this needs a clearer variable name. I don't know the context well enough to choose the best one, though.
There was a problem hiding this comment.
I agree. There is the need to rename this variable, but I don't like pstdin/pstdout either. Here some suggestions (in my decreasing order of preference):
standard_inputstdin_fdinputstdin_
There was a problem hiding this comment.
P was for process as placeholder,
I'm fine with either standard_{in,out}put or std{in,out}_fd, @eldering do you have a preference? In the end it should be clear for the person doing the debugging which will probably not be me.
If no response by wednesday I'll pick stdin_fd and merge (as its the shortest and its trivial to rename later and I need this for my next PR).
There was a problem hiding this comment.
std{in,out}_fd sounds fine to me.
judge/runpipe.cc
Outdated
| @@ -641,22 +641,22 @@ struct state_t { | |||
| tie(read_end, write_end) = make_pipe(); | |||
| logmsg(LOG_DEBUG, "setting up pipe #%ld (fd %d) -> proxy (fd %d)", i, | |||
| write_end, read_end); | |||
| process.stdout = write_end; | |||
| process.pstdout = write_end; | |||
There was a problem hiding this comment.
I agree. There is the need to rename this variable, but I don't like pstdin/pstdout either. Here some suggestions (in my decreasing order of preference):
standard_inputstdin_fdinputstdin_
vmcj
force-pushed
the
build_musl_runpipe
branch
from
d8b7a0e
to
a3a162c
Compare
Trying to build on alpine (which uses musl instead of glibc) would break as the preprocessor replaces stdin->(stdin) which fails to compile. After discussion with @ankon he found: https://lists.llvm.org/pipermail/llvm-commits/Week-of-Mon-20130506/173524.html
vmcj
force-pushed
the
build_musl_runpipe
branch
from
a3a162c
to
14ed297
Compare
Trying to build on alpine (which uses musl instead of glibc) would break as the preprocessor replaces stdin->(stdin) which fails to compile.
After discussion with @ankon he found:
https://lists.llvm.org/pipermail/llvm-commits/Week-of-Mon-20130506/173524.html