🚀 go-pugleaf

In article <Bz5pyw.CF4@unix.amherst.edu> I wrote:

>Shouldn't the posting
>mechanism post the article to the groups which *are* valid?  Should
>the x'ed groups even show up in the Newsgroups line when an article
>lands here?

I've gotten a lot of feedback on the second question, which I agree
was not well thought out.  It makes more sense to me not to rewrite
the headers, philosophically as well as technically.

However, I haven't heard much on the first question.  If inews is
handed a post with several groups on the Newsgroups line, some of
which have "x" (or "n", for that matter) in their active file entries,
why shouldn't inews simply post the message to the remaining groups
and ignore the invalid ones?  (After warning the user that some of the
specified newsgroups are disabled locally, of course.)  I can't think
of any situation in which I wouldn't want this to happen.

I'm new to news administration -- a fact which should probably come as
news to few of you -- so I apologize if this is a FAQ.

-- 
____ Tim Pierce                / 
\  / twpierce@unix.amherst.edu /               Rocks say goodbye.
 \/ (BITnet: TWPIERCE@AMHERST) / 

A general rule of good software design is that if an error is generated,
the software should abort with no side effects.

If a warning is generated, the software should continue as appropriate.
A warning in general is about something that likely has no bad consequences
or can easily be undone.   An uncasted type conversion often merits a warning
by a compiler -- the user can always fix it.

The question is deciding what's an error and what's a warning.  I think
posting to an invalid newsgroup or an unpostable group is an error.  The
user wants the chance to correct the error and repost.

Posting an article causes changes which can be cancelled, but not entirely
undone.  Thus just about any problem in posting should be an error causing
an abort and a message to the user -- making sure the article text is available
for reposting after correction.

-- 
Brad Templeton, ClariNet Communications Corp. -- Sunnyvale, CA 408/296-0366

In article <1992Dec22.220929.21666@clarinet.com> brad@clarinet.com (Brad Templeton) writes:

>The question is deciding what's an error and what's a warning.  I think
>posting to an invalid newsgroup or an unpostable group is an error.  The
>user wants the chance to correct the error and repost.

It's a stickier issue than that, though: a user who posts to an
invalid newsgroup and, simultaneously, to a valid one has just
committed an error, but has not committed an error.  Very Zen.

>Posting an article causes changes which can be cancelled, but not entirely
>undone.  Thus just about any problem in posting should be an error causing
>an abort and a message to the user -- making sure the article text is available
>for reposting after correction.

This can be fairly inconvenient for groups which have a high
percentage of crosspostings.  Perhaps the posting interface could
identify the invalid groups and, if any are present, tell the user
"The following newsgroups may not be posted to locally: alt.foo.bar,
comp.blat.zip, etc.  Post to remaining groups (y/n)?"  This would
allow the user to abort if necessary, but would reduce the hassle for
the (fairly common) case in which the user *does* wish to post to the
valid groups.  Besides, it seems like an appropriate halfway measure
for an error which is not an error.

What do y'all think?

-- 
____ Tim Pierce                / 
\  / twpierce@unix.amherst.edu /               Rocks say goodbye.
 \/ (BITnet: TWPIERCE@AMHERST) / 

I think the Newsgroups header should remain intact by default, so that
on systems which do carry the locally-Xed groups do get the
appropriate followups.  I'm not sure the poster should be asked *on
followups*, unless the posting sw can determine that the poster
*added* locally-Xed groups.

--
J Lee Jaap <J.L.Jaap@LaRC.NASA.Gov> +1 804/864-2148
    employed by, not speaking for, AS&M Inc, at
         NASA LaRC, Hampton VA 23681-0001