This is the mail archive of the gcc@gcc.gnu.org mailing list for the GCC project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]

Re: Idea, and obstacle.


On Mon, 22 Oct 2001, Gerald Pfeifer wrote:

> After pondering a bit, though, I believe that in fact WWW (gnatsweb,
> mailing list archives,...) is one of, if not the single, hardest thing
> to deal with correctly.  For the mailing list archives, for example, it
> is critical to keep old addresses as there are lots of references in
> bug reports, patches, etc.

We need to work a lot harder at this.  A reminder: the 4492 files listed
at http://gcc.gnu.org/ml/gcc/2001-08/msg01443.html are still required from
the old web disk.  The time this restoration is taking is thoroughly
excessive.

What is the current gcc.gnu.org backup schedule?  If there isn't yet a
proper system for backups (at least, daily on weekdays) and people who
have the responsibility of fully handling restoration where necessary
within days, not months, we ought to consider moving the whole project to
the FSF systems, if they do better in this regard.  For comparison, what
are the backup schedules of the FSF systems?

> 1. We could leave gcc.gnu.org untouched (at least for now) and address the
> style issues, incrementally?
> 
> 2. In general, I recommend to use DNS more fine-grained than we currently
> do, by having cvs.gcc.gnu.org, mail.gcc.gnu.org, gnats.gcc.gnu.org, and
> so forth; similarily for other projects.
> 
> That way, one can easily move single services among hosts (like CVS from
> some non-FSF CVS server to subversions, from subversions to savannah,
> etc., or just move GNATS).

It should also be possible to arrange for requests to a gcc.gnu.org IP
address to be routed to different machines depending on the requested
port.

-- 
Joseph S. Myers
jsm28@cam.ac.uk


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]