On Thu, May 8, 2008 at 12:17 AM, Tom Lane <tgl@sss.pgh.pa.us> wrote:
> Bruce Momjian <bruce@momjian.us> writes:
> > I think it would be helpful for us to provide an infrastructure where
> > people who don't run their own servers to store their patches at a
> > stable URL where they can keep updating the content. I did that with
> > the psql wrap patch and it helped me.
>
> Actually, I find that that is a truly awful habit and I wish that people
> would *not* do it that way. There are two reasons why not:
>
> * no permanent archive of the submitted patch
>
Yes. I can see how posting a URL to a patch would be convenient, but
having the permanent record of the patch as submitted is important.
What about uploading patches to the wiki? That way we have the
permanent record (change history), as well as the single authoritative
location for fetching the latest version.
> * reviewer won't know if the submitter changes the patch after he
> downloads a copy, and in fact nobody will ever know unless the submitter
> takes the time to compare the eventual commit to what he thinks the
> patch is
>
Well, as long as you send another message to the lists saying "I've
uploaded a new version of the patch, that URL again is <>". If you
just silently update the patch without telling anybody you're bound to
run into problems.
Cheers,
BJ