Re: [PATCH 2/6] DW attribute macro MACRO_AT_func and MACRO_AT_range

[Doug Evans <dje@google.com>]

On Sun, Nov 9, 2014 at 6:04 PM, Yao Qi <yao@codesourcery.com> wrote:
> I am inclined to delete them via "remote_file host delete" as many
> executables func_addr[pid].x are generated.  Supposing dwarf assembler
> works perfectly, people have only to concentrate on their test artifacts
> instead of func_addr[pid].x.

OTOH, leaving such things around for debug purposes is important,
or at least not making it difficult to keep such things (and editing source
files to, e.g., comment out the "file delete", crosses a threshold of
annoyance for me).

I think the simplicity of implementation and assistance to debugability
of just leaving the file there outweighs the extra code and any perceived
increase in cleanliness.

Plus the more remote file operations we add the more we slow
things down.  There's a *ton* of files we leave behind,
I wouldn't worry about leaving a few more behind.
If there were another reason then that might be something
to consider.

[At least with gcc there is -save-temps.  We *could* have something
like that here, but I'm ok with just leaving them.]