This is the mail archive of the
mailing list for the GCC project.
Re: Linux ia64.h
- To: obrien at FreeBSD dot org
- Subject: Re: Linux ia64.h
- From: Jim Wilson <wilson at cygnus dot com>
- Date: Mon, 14 May 2001 14:33:41 -0700
- Cc: gcc at gcc dot gnu dot org
- Newsgroups: cygnus.egcs
In article <20010512020539.A68071@dragon.nuxi.com> you write:
>1. Shouldn't gcc/config/ia64.h have a copyright header?
The config/ia64/ia64.h file does have a copyright header. The config/ia64/
linux.h file does not currently have a copyright header.
Not all files are required to have copyright headers. If a file is small
enough, it does not need a copyright header. There is no point in putting
a 20 line copyright message in a file with only a couple of lines of code.
When the config/ia64/linux.h file was created, it was not big enough to
justify a copyright header. It has grown since then, and is perhaps big
enough now that it should have one, but this is a difficult judgement call,
and is something that is easy to miss unless someone points it out. Just
to make this more confusing, more than half the stuff in ia64/linux.h should
actually be in ia64/sysv4.h (I didn't have enough info at the time to tell
where to put some stuff), and after this clean up, the file would likely be
small enough again that it wouldn't justify having its own copyright header.
>2. Why is this new header allowed to include other config headers
> directly?? I was prevented from committing FreeBSD headers that did
> this. Why was I being held up to a different standard than any one
> else is?
You aren't being held to a different standard. We have are old ports that
haven't been cleaned up yet, and occasionally some spotty enforcement of
the rules. There is also the question here of what "new" means. The
ia64/linux.h file is over 15 months old in the FSF tree. If you consider
the NDA Trillian tree, then the file is over 2 years old. When I created the
file over 2 years ago, I followed then existing common conventions, and the
file has not yet been updated to match all current conventions.
The rule is a good one, and just because there are some ports that don't
conform to it does not mean that we should throw it away.