Subj : Release of v3.4?
To   : Kees van Eeten
From : Roy Witt
Date : Fri May 24 2013 10:01:50

Kees van Eeten wrote to Roy Witt:

 RW>>>> My RC, Marc Lewis can't use MNL on his OS/2 system and has
 RW>>>> reverted back to using NLMake. Some of the new keywords produce
 RW>>>> errors when MNL encounters a PVT listing with a 000.0.0.0.0 number
 RW>>>> in the phone field of an ION node. We've been waiting for months
 RW>>>> on this issue, which Andrew Leary has been aware that it exists
 RW>>>> and has stated that the MNL team is working on it.

 KvE>>>  Someone must have missed something in the specs.

 RW>> That could be. My MNL compiled netseg crashed the RC19's OS/2 system
 RW>> and so did a few other netsegs in R19...meanwhile, I've blocked the
 RW>> keywords that crashed his system...


 RW>> ; control file for Net 387 using
 RW>> ; MakeNL v3.3.1 -- Windows 32 Bit Version
 RW>> ; nodelist control file.
 RW>> ;ALLOWUNPUB 1
 RW>> ;MINphone 1

 RW>>   ________ Roy snipped text here!________O/___________________
 RW>>                                          O\

 KvE>  I have looked at your remarks again. It seems to me that you talk
 KvE> of two distinct problems.

 KvE>  The first is where you get error messages is a specific case, the
 KvE> seconds where your RC has a program crash on a OS/2 system.

Actually, I get no error messages as MNL works as advertised on my WinXP
machine.

 KvE>  In your case a line with  Pvt,......,000.0.0.0.0,........
 KvE>  produces an error message.

This is the line that crashes OS/2 on my RC's system.

,10,CompTA_Radical,San_Antonio_TX,Roy_Witt,-Unpublished-,300,CM,MO,XX,INA:
z28krazy.dyndns.org,IBN

This is the LOG file that MNL created that shows a succesful compile here.

- 11-Jan-2013 11:22:20 makenl[204] MakeNL 3.3.1 (Win32) compiled with
Watcom C on Dec 26 2012 00:19:36
+ 11-Jan-2013 11:22:20 makenl[204] MakeNL started
+ 11-Jan-2013 11:22:20 makenl[204] Cmdline: E:\MNL\3.3.1\makenl.exe
"n387.ctl"
+ 11-Jan-2013 11:22:20 makenl[204] Using 'n387.ctl' in 'E:\MNL\3.3.1'
+ 11-Jan-2013 11:22:20 makenl[204] Begin processing '387net.' -- 11:22,
Friday, January 11, 2013
+ 11-Jan-2013 11:22:20 makenl[204] Sending 'E:\DB3\NETMAIL\387net.011' to
19/0
+ 11-Jan-2013 11:22:20 makenl[204] CRC = 38474
+ 11-Jan-2013 11:22:20 makenl[204] MakeNL finished (rc=0)

BTW, as the date of the log indicates, this is how long it has been since
the problem with OS/2 has been known to exist.

 KvE>  Your current segment does not list a dotted number in te telephone
 KvE> field.

Correct. But if you look at node 387/10, that is the line that crashed
OS/2 which was formerly not PVT and had -Unpublished- in the phone
field...

 KvE>  And the logic in the programs is that a line starting with Pvt, has
 KvE> the
 KvE>  word "Unpublished" in the telephonenumber field.

The latest MNL program is supposed to allow an ION node to use
-Unpublished- in the phone field without having to use the PVT flag.

 KvE>  The flag "ALLOWUNPUB 1" allows the word "Unpublished" to be used in
 KvE> lines
 KvE>  starting with other tags than "Pvt"

Correct. But, using that flag in MNL crashes OS/2...

 KvE>  The flag "MINPHONE 1" counts the number of "-" separated groups in
 KvE> a telephone number. For a value other than "Unpublished" it should
 KvE> be set to "1" if there are no dashes in the phone number field.

Yes, see above.

 KvE>  What is not clear what causes makenl to crash on OS/2. If it is
 KvE> something not in line with the specs, it should produce an error
 KvE> message, so it is something more serious.

I do not have an OS/2 machine to varify that.

 KvE>  If those who have this problem are not in a position to hunt for
 KvE> the problem themselves, someone else has to be found, who can
 KvE> reproduce and analyse the same problem.

I'd volunteer my RC, but he's probably just getting out of surgery today
and not up to the task...

 KvE>  Lets hope that we can entice Ulrich to have a go, or whoever
 KvE> compiled the executables for OS/2.

?

 KvE>  As a last thought, there should not be a problem in using NLMAKE,
 KvE> if it can handle the envisioned changes to the nodelist.

Apparently it does have a problem as it doesn't like the segment without a
PVT flag and -Unpublished- in the phone field either. That is why the two
keywords are commented out in my segment and node 10 is flying PVT and 000
etc. numbers in the phone field. Once MNL can work with an OS/2 system, I
can uncomment those two keywords, revise the node 10 line and forward my
revised segment to my RC...


         R\%/itt


--- GoldED+/W32 1.1.5-31012
--- D'Bridge 3.92
 * Origin: Lone-Star BBS - San Antonio, Texas - USA (1:387/22)

.