Text 1499, 83 rader
Skriven 2008-03-25 00:28:22 av mark lewis (1:3634/12.0)
   Kommentar till text 1436 av Alexey Vissarionov (2:5020/545)
Ärende: Proper nodelist format for Internet-only nodes?
=======================================================
 CS> The way the software works, it has to be a top down change. It's 
 CS> like when we put in the bit to allow above 9600 speeds to be 
 CS> listed. If an NC did it and the RC wasnt set, the NC's segment 
 CS> update failed.

 AV> That would be the RC's fault.

say what? fidonet is a top-down structure... you can't do things that affect
the whole network from the bottom and expect them to make it to the top... to
do so could very easily shut down the entire network... wasn't it you that
asked me about security and such as this?? it would be rightly stopped at the
RC's system so that it couldn't propogate further on and cause problems with
the rest of the network...

 CS> Janis may or may not have coordinated the shift with the other ZC's
 CS> yet or it may be in progress. I don't know. Once she's ready (or in
 CS> your case it would be Ward), notification goes the the RC's to shift
 CS> then each RC tells their NC's when they can start using it.

 AV> Incorrect. The decent solution would be accepting both syntax
 AV> during the transition period.

at what level? remember, fidonet is not a single level network...

 CS> Come to think of it, this might be part of why Ward was doing all
 CS> those RC checks? Not sure, he didnt mention it that I saw (could have
 CS> just been time to check again).

 AV> Perhaps using the validator written in Perl will be a good
 AV> solution... 

[sigh] yet another proposed non-FTN solution... there are already validators
written that are already FTN compliant... why do you keep wanting to reinvent
the wheel? we've already got several that are round, support the structure,
carry the weight, and are easily used by the majority of systems :?

 CS> Anyways if an NC files the new way when the RC or ZC above them is stil
 CS> lon the older software or hasnt changed the settings to allow the
 CS> '-Unpublished- without a PVT at the front, the segment will fail
 CS> upstream.

 AV> That will be the upper *C's fault.

ahem... the shit does not roll uphill in fidonet... or anywhere else, for that
matter :)

[trim]

 >>  CS> Implict though in the Z1 method is 000-0-0-0-0 means
 >> '-Unpublished-'.
 >> It does not comply with FTS-5000 and, even worse, it may be
 >> misinterpreted.
 CS> It is not documented in the FTSC series, correct.

 AV> Incorrect. It is neither valid phone number nor valid IP address. 

it is not supposed to be a valid ip address! it is a uniform string of
characters that can be easily detected and acted upon... that's what it is
intended to be used for, too... would you rather try to code a detection for
the myriad of other strings that were used before anyone standardized this one?
i don't think so... IIRC, one tool that i wrote had to contend with looking for
some 20 variations, at least... now there's only one to look for and you and
others want to bitch and whine about it :(

 >>  CS> It can be safely globally edited to that if needed.
 >> Running `sed -i -re 's/,0[0\-]+0,/,-Unpublished-,/' nodelist.ndl` is
 >> quite simple. We all know (from the recent discussion in FTSC area)
 >> that the only incompatible software is the ancient nodelist generator,
 >> and then it was presented as a strong reason. Now it is not a big deal,
 >> as there is the alternative:
 >> ftp://happy.kiev.ua/pub/fidosoft/nodelist/nlcheck.zip - it is written
 >> in Perl and can be easily reconfigured for any FTS-5000 updates.
 CS> We know there is a more current one and that the 000-0-0-0-0 entries
 CS> will go away (hopefully soon!) but for reasons as above listed, has to
 CS> be an organized approach.

 AV> Good reason to rework the FTS-5000...

*AFTER* things are in =widespread use=...

)\/(ark

 * Origin:  (1:3634/12)