Re: versions interoperability (Re: IRC 2.8.21 compilation)

["Rico \"mc\" Gloeckner" <mc@xxxxxxxxxxx>]

bluenet is run by ibm workers in their worktime. They cant spend hours
compiling a whole system and follow every upgrade. I think thats mostly
the reason why they dont upgrade.

Never touch a running system when you dont have the time to play with the
probably resulting trouble.

--mc
(Rico Gloeckner  [*mc@xxxxxxxxxxxxxxxxx(Newnet)] [mc-@iRCNet])
// mc@xxxxxxxxxxx // mc@xxxxxxxxxxxxxxxxxxxxxxxxx // mc@xxxxxxxxxxx //
// Fido: Rico Gloeckner@2:2444/1112.16 // ZC: ukeer@xxxxxxxxxxxxxxx //

On Wed, 10 Nov 1999, Christophe Kalt wrote:

> On Nov 10, Graham_Gillies@xxxxxxxxxxx wrote:
> |      I'm trying to compile the 2.8.21 IRCD.
> 
> pre 2.9.x version discussion isn't considered to be appropriate
> for this list, however, i think the compatibility issues are
> interesting.
> 
> |      The reason i'm using such a back level version is that the other IRC
> | servers in the network are all 2.8 servers.
> 
> backward compatibility is somewhat tricky with IRC, and shows
> a lot about the protocol shortcomings.  Basically, old and
> recent versions cannot coexist peacefully, and the realities
> of this have only been somewhat documented recently.
> 
> + 2.9.3 (and later) cannot coexist on a network with 2.7
>   servers (i forget why)
> + 2.9.4 (and later) no longer support 2.8 server links
> + most of 2.10.x new features are unavailable outside
>   contiguous 2.10.x clouds in a network
> 
> I know there are many networks using rather ancient versions
> of the IRC server software, and this is a problem for them as
> there is no longer any supported software for these.
> Basically, if you run a 5 year old irc server, you may very
> well need to run it on a 5 year old OS and figure out things
> mosly on your own.
> 
> i don't really know of any reason why someone would refuse to
> upgrade server versions, but would be curious to hear of such
> motives if they exist.  Really, i think the best way is to
> upgrade the servers progressivly.
> 
> Christophe
>