Re: [patch] General info in AS listings

[Nick Clifton <nickc at redhat dot com>]

Hi Daniel,

> > The next thing to consider is whether the feature is actually necessary ?  
> > For example it seems to me that all of the information generated by your 
> > new option could also be obtained from the command line or via a wrapper 
> > script.  So why add the option to the assembler ?  The point here being 
> > that it is a bad idea to add unnecessary options - feature bloat - because 
> > it just makes room for more bugs.
>
> I think it's useful.  Identifying information is a big aid to
> reproducibility, both for toolchain support and file archeology; both
> of which I spend a lot of time doing :-)

I can see that, but I also think that everything that Santiago's patch 
currently does can be achieved by using a wrapper script or a makefile 
rule to prepend the information to the assembler's listing output.  So 
why modify the assembler ? (*)

That said I am not actually against the patch.  I would just like to 
have Santiago defend it (or rather, include a justification for the 
feature when he resubmits his patch), and to have some contributors say 
that they like it.  I am conscious of the evils of feature bloat and I 
want to be sure that the feature is actually going to be of use to 
developers before agreeing to its inclusion.

Cheers
  Nick

(*) Answer: because putting the code into the assembler means that the 
code only has to be written once.  Using a script or makefile rule means 
that everyone who wants this feature has to modify their build 
environment, and unless there is a repository for the script/build rule 
then there will be hundreds of different versions of it, with hundreds 
of different opportunities for bugs.