6 Sep
2020
6 Sep
'20
3:39 a.m.
I think you got it.
I did a TELL MIM:: SHOW KNOWN NODES (which did not fail with a read
error this time) and compared that output with the node database on
VENTI2, which is updated weekly from MIM::T20.FIX.
It's the same.
Stupid buffers...
On 9/4/20 7:16 PM, Johnny Billquist wrote:
And I think I fixed that one as well... (Stupid
code...)
? Johnny
On 2020-09-05 01:00, Johnny Billquist wrote:
And now it seems MIM is giving even worse data.
Looked good for a
while. Sigh. Oh well, more debugging coming up. I guess you should
again regard the information with suspicion then, until a new notice
comes from me on this topic.
?? Johnny
On 2020-09-04 20:35, Johnny Billquist wrote:
I forgot. The version in M+ V4.6 is
"V08.00". You can check this
with "TAS NMVACP".
The fixed version is "V08.20".
?? Johnny
On 2020-09-04 20:34, Johnny Billquist wrote:
Happy to announce that I have found and fixed the
bug in the DECnet
code in RSX that could cause operations like "SHOW KNOWN NODES" to
skip some nodes under some circumstances.
So anyone getting node definitions from MIM can now safely do that
using NCP (in VMS), or NNS (in RSX) and be sure they do get the
complete list.
For anyone using RSX, this is another component you want to update.
The file in question is [5,54]NMVACP.TSK, and you can copy it from
MIM::
Place it in that directory, and you're good after the next reboot.
If you want it fixed on your running system, then after copying the
file over, do the following two lines:
REM NMVACP
INS LB:[5,54]NMVACP
and then you are good.
If anyone wants to know itty gritty details on the bug, let me know
and I can explain it. But suffice to say for now, that this bug was
introduced with the nodename server in RSX, which I think only came
in M+ V4.6.
?? Johnny
On 2020-08-27 02:16, Johnny Billquist wrote:
> On 2020-08-27 01:47, Dave McGuire wrote:
>> On 8/26/20 4:40 PM, Johnny Billquist wrote:
>>> Let me put it this way. If any tool causes you any problems,
>>> first check if there is something newer on MIM, and if so - grab
>>> and test that.
>>> If not, let me know, and I can probably fix it.
>>
>> ?? I'm sure I speak for all of us here when I say THANK YOU for
>> this.
>
> The thing is - some of the bugs and issues I've worked on are
> rather obscure. You might occasionally observe there is something
> funny going on, but understanding where the problem is, and
> finding the fixed file on MIM is not always so easy, nor how to
> apply it.
>
> Take MCR, for example. There are multiple things I've fixed. Both
> in MCR... and ...MCR (they are two different things). And applying
> them needs that you update things through VMR. Can't be done on a
> running system. (Well, ...MCR can, but MCR... cannot.)
>
> My TCP/IP package already does that patch for you however, so you
> should be good.
>
> Hey, here is something for people to look at, to make you aware of
> things you might be interested in: MIM::US:[BQT]FIX.TXT contains
> all the work I've been doing for a possible future V5.0 release of
> RSX-11M-PLUS.
> A lot of those fixes apply just fine also for V4.6, but don't ask
> about the subdirectory stuff. That is just too complex to patch
> up, and there are still a couple of corners I'm not done with.
>
>>> And while I'm on it. At the moment there is a long standing bug
>>> in NICE on RSX, which have been annoying me slightly for a long
>>> time, but which is now starting to become serious enough that I
>>> need to look deeper into it. If you do a "NCP TELL MIM SHOW
>>> KNOWN NODES" the list is incomplete. If you look carefully, some
>>> nodes are missing, which will be there, if you explicitly ask
>>> for them.
>>>
>>> This becomes a larger problem because if someone on a VMS system
>>> does an "NCP COPY KNOWN NODES FROM MIM ..." they will not be
>>> getting the full list either. And the same goes for RSX users
>>> using NNC to copy the known nodes from MIM.
>>
>> ?? Ah-HA!!? I have seen exactly this.? I thought I was losing my
>> mind.
>
> Well, now you know you aren't. :-)
>
> ?? Johnny
>