[net.news.config] Responses to netnews articles mailed from 'vnews' lost!

pcl@ihnp4.UUCP (Paul C. Lustgarten) (02/11/84)

We have just discovered that our version of vnews has been quietly
throwing away ALL replies - i.e., the 'r' command is broken.  The
observed action is that an 'r' command puts you in your EDITOR, you
enter your reply, and on exiting, vnews continues as expected - and as
though the reply you just composed had been sent.  It instead never even
made it to /bin/mail.

This condition has presumably been in effect for several months, and
(apparently) applies to all netnews sites at Bell Labs' Indian Hill and
Short Hills locations (about 80 sites - ih*, iw*, ix*, and hl*).

Since we have not yet tracked down the cause of this misbehavior, we
don't know if it is a purely local bug or not.  We encourage all other
sites running 'vnews' to confirm that their replies are indeed being
sent.

	Paul Lustgarten
	AT&T Bell Laboratories, Short Hill, NJ
	ihnp4!pcl	ihnp4!hlexa!pcl

bbanerje@sjuvax.UUCP (02/15/84)

>> We have just discovered that our version of vnews has been quietly
>> throwing away ALL replies - i.e., the 'r' command is broken.  The
>> observed action is that an 'r' command puts you in your EDITOR, you
>> enter your reply, and on exiting, vnews continues as expected - and as
>> though the reply you just composed had been sent.  It instead never even
>> made it to /bin/mail.
>> 
>> This condition has presumably been in effect for several months, and
>> (apparently) applies to all netnews sites at Bell Labs' Indian Hill and
>> Short Hills locations (about 80 sites - ih*, iw*, ix*, and hl*).
>> 
>> Since we have not yet tracked down the cause of this misbehavior, we
>> don't know if it is a purely local bug or not.  We encourage all other
>> sites running 'vnews' to confirm that their replies are indeed being
>> sent.

We had the same problem.  I finally fixed it, though I don't know
why the fix worked.  Try changing the '(' parentheses to '{' parentheses
in the 'reply' and 'follow' shells.  This of course applies to the
matching braces as well.

The '(' serves to fork a sub-shell while the '{' merely groups the
commands.  If this doesn't work, check to see if you are using the
correct version of recmail.  The one that comes with the news 
distribution doesn't work.  You have to use the one that came with
vnews.

I would have replied by mail, but I am interested in whether anyone
out there has the answers as to why the parentheses should be 
important.  I crave the indulgence of those who read this group.
-- 


				Binayak Banerjee
		{allegra | astrovax | bpa | burdvax}!sjuvax!bbanerje

wls@astrovax.UUCP (William L. Sebok) (02/17/84)

Another thing which may or may not be relevant.  Sometime last fall I posted
a suggested bugfix to the reply and follow shell scripts. A ^Z stop signal
given in vnews after a reply will cause the mailer or even the uucico that the
mailer spawned to stop with possible nasty consequences (readnews also has this
bug).  This bugfix involved having the shell script add signals 18, 19, and 21
to the list of signals to be ignored.  Unfortunately in order for this to work
"sh" has to be recompiled with MAXTRAP in defs.h set to 32 rather than 17, to
let the shell recognize the existence of these signals (I regard this as a
bugfix to sh.  I don't understand why the shell as distributed with 4.X BSD
doesn't recognize that these signals exist.  Sure, sh doesn't want to do full
job control but it could at least allow one to ignore the job control signals).

Therefore if by some chance one has been using reply of followup shell scripts
with my modifications but have not made the necessary changes to sh, then
replies will be lost.  This broke vnews at many sites (mine included) when the
shift from BSD 4.1 to BSD 4.2 was made because the sh distributed with 4.2
did not include the necessary change.

This has been a long hiatus since my last posting.  After several months
working on the 4.1 --> 4.2 conversion at astrovax I have it about 60% back
to working as well as it did under 4.1.
-- 
Bill Sebok			Princeton University, Astrophysics
{allegra,akgua,burl,cbosgd,decvax,ihnp4,kpno,princeton,vax135}!astrovax!wls