PE>> And there could equally well be a problem with PQWK221, you wouldn't
 PE>> know.

 RS> However I have been using 2.02 for a very long time and if it does have
 RS> another problem that I am not aware of, it must be one which is triggered
 RS> by some obscure combination or by something I dont actually do. Yes, it
 RS> would be nice if I could be sure it was perfect, no one has actually
 RS> invented a way to do that tho. But its quite straight forward to test say
 RS> 2.22 carefully by comparing the PKT it produces with what 2.02 produces
 RS> to see that what difference there are are acceptible. Which happens to be
 RS> what I plan to do.
 
I was talking about Frank and 2.21, not 2.02. 

 RS> Notice you studiously deleted the examples where inadequate testing
 RS> in the past has resulted in fangs in the arse.

I can't remember what that was about, but my point is there is
just as likely to be problems in 2.21 as 2.22.

 PE>> They are ALL tested like that,

 RS> Yes, and it obviously didnt not pick those problems that go thru up.

Yep, and just as likely to be in 2.21.

 PE>> if Frank is waiting for any long-term bugs to appear, they are
 PE>> just as likely to be in PQWK221 as PQWK222

 RS> There is the tiny matter that he has actually been running whatever
 RS> he has been running for a while already.

The difference between a month and a week is irrelevant as far as
long term problems are concerned.  Coupled with the fact that there
is a very real problem with 2.21, which is KNOWN, whilst there are
no KNOWN problems (ie short-term) with 2.22, makes it highly
desirable to switch to 2.22.

 RS> Your nutty theory now implys that better testing is pointless. Thats crap.

Eh?  I don't imply anything of the sort, because I don't
claim/believe that at all.

 PE>> (in fact if they are in one they will almost certainly be
 PE>> in the other, since I only changed a couple of lines of code).

 RS> Nice theory Paul. Didnt explain the intrazone netmail breaking in
 RS> an attempt to fix the inter zone netmail tho. And it would not be the
 RS> first time in recorded history that an attempt to fix one problem has
 RS> broken something else either. Which just happens to why better testing
 RS> is a good idea on stuff that matters, particularly when the result of
 RS> something breaking may not be visible.

Of course, that's why Brenton tested it out first.  And it
works.  And when there's a major problem with 2.21, and a
particular user who is very prone to discovering it every
single bloody netmail he sends, it makes a lot of sense to
switch to what three other people have been using without
problems.  I already suspect 2.22 has a bug in it - the same
one that is in 2.21 with regards to taking a long time to
process humungous packets.  I'm not saying 2.22 is perfect,
just that it is a far better alternative than 2.21, and
that there is no valid reason to be sticking to 2.21.  
BFN.  Paul.
@EOT:

--- Mksmsg