MvdV>> when the INA flag came into the picture over a decade ago, I MvdV>> proposed having protocol flags inherit information from those MvdV>> preceding them as an alternative.
ml> i remember... the problem with that is how to indicate a "no ml> capability"...
Eh? I do not understand. What do you mean by "indicating no capability"?
ml> in the rewrite of the example i gave, you ended up listing the final ml> backup system as the main connection and that is not how that ml> connection's entity was designed...
You can change the order but it will not make any difference:
The meaning of the flag is indpendent of its position.
ml> it would be like listing three MXes and having everyone send to the ml> last backup instead of to the main one...
Wrong analogy. MXs carry a priority. There is no equivalent for that with IP connect information in the nodelist. There is possibility to list multiple addresses by repeating a flag with a different host name or IP address, but that's it. There is no implied priority by order. Wether the application uses the first, the last, tries them all or alternates in round robin fashion between sessions, is up to the implementation.
ml> yes, the example given was on the ""extreme"" end of the ml> belt but it showed how explicitly listing and ordering of the flags ml> can be used so that one can set their systems they way they want/need
Sorry, but that is not how the system evolved. Flag meaning is independant of flag order. It is not what I would have choosen, had the decision be mine, but it was not. When introducing the INA flag, always listing it before the associated protocol flags would have been a good start. But the nodelist clerks in their infinite wisdom decided otherwise and now there is no way back. We just have to live with that.
ml> them to operate instead of being forced to operate in the way that ml> others tell them to operate... many drop out because of things like ml> this that can't be resolved to suit their needs...
Then perhaps those people had there expectations set too high. When interconnecting nodes in a network one has to abide by a set of common rules to make it work. Adapting the rules when the network evolves is fine, but one has to realise that one can not expect the whole network to bend over backwards to instantly accomodate every demanding nerd's need. Real or perceived. Cooperation is the key word to the smooth operation of the network and cooperation is not a one way street. It also means that one can not always get what one wants. If one can not accept that one can ask too much, that what one wants, is too outlandish, puts an unreasonable demand on the network, or even worse conflicts with the justifiable needs of others, then they will just have to choose between adapting or leaving...
MvdV>> That was voted out/shouted down/ignored as well. Instead the MvdV>> system continued into one where flag meaning does not depend on MvdV>> order.
ml> i never saw that... what i saw was positional usage which was promoted ml> more with the multiple INA flags capabilities...
The opportunity to introduce positional usage went out of the window when the nodelist clerks allowed listing the INA flag in an arbitrary position instead of only before the associated protocol flags. You should have spoken then. Now it is too late.
MvdV>> That happens, one can not always get the system to adopt what MvdV>> one thinks is the best way.
