[Freeciv-Dev] (PR#8497) loading transporters: behavior of different commands

["Jason Short" <jdorje@xxxxxxxxxxxxxxxxxxxxx>]

<URL: http://rt.freeciv.org/Ticket/Display.html?id=8497 >

Currently a unit is loaded onto a transporter when:

1.  The load command is invoked explicitly.  (This may also sentry the 
unit.)

2.  The unit is sentried on the same tile as a transporter it can be 
loaded onto.  (Currently this is done automatically by the server, but 
it should be done by the client instead.)

3.  The unit moves onto the same tile as a transporter it wants to be 
loaded onto (e.g., airplanes auto-land on carriers).  (This is done 
automatically by the server; it *could* be done by the client instead.)


My question is, what *should* the behaviour be?  IMO:

1.  When the load command is invoked, the unit should be loaded.  It is 
also sentried if the sentry_on_load client option is set.

2.  When a unit is sentried it is _not_ loaded.  This allows units to be 
safely sentried in cities.  If the player wants to load a unit he uses 
"load" instead.

3.  When the unit moves onto the same tile as a transporter it wants to 
be loaded onto.



A unit is *un*loaded from a transporter when:

1.  The unload command is invoked explicitly.  (This is hard to do since 
the unit must be activated already to invoke it; see #3.)

2.  The unit tries to move.  (This has to be done by the server since 
the unload is done as a part of the move.)

3.  The unit is activated while at a location where it doesn't "want" a 
transporter (this means warriors in a city will be unloaded but not 
airplanes at sea).  (This is currently done by the server but should be 
done by the client.)


Again, what *should* the behavior be?  IMO:

1.  When the load command is invoked, the unit is unloaded (if possible).

2.  When the unit tries to move, it is unloaded (if possible).

3.  When the unit is activated, it is *not* unloaded.  To unload a land 
unit in a city the player must select the warriors and choose the unload 
command.  Perhaps a client option unload_on_activation would control 
this instead.


I expect lots of disagreement on this.

jason