lli -mcjit-remote-process

[Kaylor, Andrew]

I'm not sure where the first comment you reference came from.  I've cleaned that up.

As for the option name, I'm not particularly attached to the name it has, but it doesn't seem to me that renaming it offers significant improvement.  This option is only there for testing purposes.  I wouldn't oppose a patch to rename it, but I don't think it's worth the effort.

-Andy

-----Original Message-----
From: Alp Toker [mailto:alp at nuanti.com] 
Sent: Monday, October 28, 2013 4:20 PM
To: Kaylor, Andrew
Cc: llvm-commits
Subject: lli -mcjit-remote-process

Hi Andrew,

In r191843 you added the -mcjit-remote-process flag:
>
>   // Manually specify the child process for remote execution. This 
> overrides
>   // the simulated remote execution that allocates address space for child
>   // execution. The child process resides in the disk and communicates 
> with lli

By definition a process resides in memory, not on disk. Is there a terminology problem in this commit?

>   // via stdin/stdout pipes.
>   cl::opt<std::string>
>   MCJITRemoteProcess("mcjit-remote-process",
>             cl::desc("Specify the filename of the process to launch "
>                      "for remote MCJIT execution.  If none is specified,"
>                      "\n\tremote execution will be simulated 
> in-process."),
>             cl::value_desc("filename"),
>             cl::init(""));

A process is something that's already running and would usually be referred to with a PID number in a flag, but the flag is documented as accepting a filename of something that is to be run.

I feel this needs a better name along the lines of -mcjit-executable plus a tidy up to the documentation and comments to reflect that.

Alp.

--
http://www.nuanti.com
the browser experts