From: John Stanley (stanley@peak.org)
Date: Fri Mar 01 2002 - 12:49:01 CST
Russ Allbery (rra@stanford.edu):
>Your rewriting of my
>message to add a degree of strictness, concreteness, and insistance which
>was not present in the original message was not helpful.
Your argument is in favor of looseness and permissibility, not strictness
and insistance, and I've always presented it as such. You were arguing
that a SHOULD NOT was too strict, and I did not "rewrite" your argument to
say otherwise. In fact, it was pretty clear that I was saying you were
trying to be too loose and for all the wrong reasons.
>medium. As such, when they occur to me, I raise issues that come out of
>use of netnews for things other than world-wide discussion groups, since
>otherwise those issues may not be raised in this forum.
Gee, wouldn't it be too bad if issues of how to turn news injectors into
bug ... oops, "request tracking systems" weren't raised in the formum
intended for a news standard. And gee, maybe others here have some clue
about using news for other than simple discussion, too.
>The primary purpose of a standard is not to create a "right to complain,"
Now who is paraphrasing incorrectly? Did I say what the primary purpose of
a standard was? Did I even MENTION the primary purpose of a standard?
No, I told you that by arguing that SHOULD NOT was too strict for "change
any header..." you lose the right to later argue that SHOULD NOT is not
strict enough when you find out that you don't like what people are doing.
>The
>primary purpose of a standard is to allow people to write interoperable
>software.
Yes. In this case, news software. Not bug ... oops again, "request
tracking software" or IRC clients. Yet that is what we should consider
when we discuss a prohibition on inserting or changing a sender header.
>If you write software that will break if the injector modifies
>any header other than those listed, there are some news servers that will
>break it.
You do understand that a standard can specify mandatory behaviour not
because some software somewhere might break if it didn't, but so that
everyone will be working from the same definitions and can expect the same
behaviour from conformant software, don't you? As in, nothing breaks if a
References header is left off a followup, at least, nothing written well
will, yet that header is a MUST for a followup and a MUST NOT for
not-a-followup, according to us. (Except, as I recall, for a proposed
moderated version of news.groups, where it was not mandatory.) And we
define the meaning of From and Sender headers so everyone uses the same
definition, unless, of course, someone wants to change the meaning as
local site policy which we, of course, cannot prohibit in conforming
applications, and the user would have no way of knowing about until too
late. Of course.
>There are some news servers that rewrite the From header, for
>example.
Another good example for a MUST NOT. Do you have some excuse for why they
should? As in, maybe they are going to be used as an FTP server by
somebody? Did they accept the header as valid? If so, where do they get
the idea it needs to be changed? If not, why didn't they reject the
article in the first place? What's this "middle ground" nonsense of
accepting it as valid now and then changing it later?
>SHOULD means that you have to be prepared to deal with it even if it's not
>supposed to happen,
Ummm, no, SHOULD means you have to be prepared to deal with it because it
IS supposed to happen.
You said:
>I guess it depends a lot on what medium and set of users you see our
>standard as addressing.
Yes, and I made the point that the medium was news and the set of users is
news users, not bug, damn, request tracking systems or mail agents, and
you dissmissed that point offhand because I used the word "bug" instead of
"request". Well, the point is still here. Don't argue that bug tracking
systems would be hindered if we say SHOULD NOT unless you are arguing that
we should not say SHOULD NOT, because otherwise it doesn't matter.
And you know, I let your first remarks about how news software should
consider non-news apps go by uncommented. It wasn't until you came up with
a better example of a non-news app that would not be compliant with a
news standard if they couldn't change headers at will that I pointed out
the problem with your argument. So it wasn't just once you brought the
issue up, it was when you decided to refine the argument to make it more
persuasive.
And guess what, I already told you, you won. Two votes for removing
"SHOULD NOT change any header". Make it a free-for-all.