To PR Senders

[Mark Linimon]

On Sat, 21 Aug 2004, Ceri Davies wrote:

> Pulling down a patch via fetch or similar is basically a PITA when they
> need follow ups in the PR, and I would rather see patches in GNATS for
> pretty much everything.

I'd have to agree -- the key item being that the more eyes that see it
initially, the more people will probably comment on it/submit patches.
However, if something is of very limited interest and is large (200k
lines of whitespace patches), then it should probably be in a URL.

On checking, it seems that most ports PRs are <20k.  A few are huge,
and sometimes that's because someone attaches (e.g.) a config.log.
I'd rather see huge files like that, that may be very specific to
one installation and possibly even transient, not be in GNATS.
However, kernel tracebacks do need to be included for the kern PRs
to be useful; but 'syncing disks 1 1 1 ...' is just wasted space.

We should probably come up with a hard limit above which you should
not put it into GNATS.

mcl