Quoting pitermagram <pitermagram@...> from ml.gnutella.dev-forum:
:Hi,
:
:i have the following question about "UP" ggep:
:
:in LimeWire client sources when this ggep parsed/created first comes ultrapeer
:version, then amount of free leaf slots and third comes amount of free
ultrapeer
:slots:
:(see attached part of source code)
:payload[0] = 0;
:payload[1] = localPongInfo.getNumFreeLimeWireLeafSlots();
:payload[2] = localPongInfo.getNumFreeLimeWireNonLeafSlots();
:
:while in GTK Gnutella client first comes ultrapeer version, then amount of free
:ultrapeer slots and third comes amount of free leaf slots (NOTE: ultrapeer and
:leaf slots are switched):
:(see attached part of source code)
:ggep_stream_write(&gs, &meta->version_up, 1) &&
:ggep_stream_write(&gs, &meta->up_slots, 1) &&
:ggep_stream_write(&gs, &meta->leaf_slots, 1) &&
:
:Is there any reason for this ?

Yes, the reason is that gtk-gnutella follows the published specifications!
The following page:

http://wiki.limewire.org/index.php?title=Known_GGEP_Extension_Blocks

defines the "UP" extension payload as:

{VersionGuessStyle VersionUP, byte SlotsFreeUP, byte SlotsFreeLeaf}

Note that the specification does not tell us what "versionUP" means, hence
this is open to interpretation. Gtk-gnutella uses the minor version of
the QRP protocol, i.e. 2 (as it supports 0.2).

Raphael