<?xml version="1.0" encoding="utf-8"?>
<rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom"><channel><title>Recent changes to bugs</title><link>https://sourceforge.net/p/xconq/bugs/</link><description>Recent changes to bugs</description><atom:link href="https://sourceforge.net/p/xconq/bugs/feed.rss" rel="self"/><language>en</language><lastBuildDate>Fri, 23 Mar 2007 11:28:31 -0000</lastBuildDate><atom:link href="https://sourceforge.net/p/xconq/bugs/feed.rss" rel="self" type="application/rss+xml"/><item><title>mp-to-enter-unit 0 not implemented correctly</title><link>https://sourceforge.net/p/xconq/bugs/121/</link><description>mp-to-enter-unit 0 is not implemented correctly:
units are charged at least 1 mp to enter a transport
E.g., in the standard game, an armor uses 2 mps
to enter a nearby town, instead of 1.</description><dc:creator xmlns:dc="http://purl.org/dc/elements/1.1/">Massimo Campostrini</dc:creator><pubDate>Fri, 23 Mar 2007 11:28:31 -0000</pubDate><guid isPermaLink="false">https://sourceforge.net5b91c9f124af0f06c63267f222d96e9f2b27afbd</guid></item><item><title>Can't build in own location</title><link>https://sourceforge.net/p/xconq/bugs/120/</link><description>Visible in the standard game with infantry building bases, but 
seems to be universal:  units cannot build in their own 
current cell.  Move an infantry unit to an otherwise-empty cell, 
do "build", choose the unit's own cell, and nothing happens; 
no error message, but the unit remains selected, waiting for 
an order.  I'm flagging this as a Tcl/Tk interface bug because 
that's where I suspect the problem is; I note that the AI 
seems to have no problem building bases, so I think what's 
going on is that the interface is somehow deciding that the 
order is illegal, and blocking it before it gets to the engine. 

It is true that an infantry unit cannot be in a cell \(in the cell as 
opposed to as an occupant of something\) when the cell also 
contains a base, but in the past the behaviour has been that 
the infantry unit can still create a base, and it becomes the 
occupant of the newly-created base immediately.  That still 
happens if you specify a location to buildother than the unit's 
current cell - it moves to the new cell, then creates an 
incomplete base, and immediately occupies it.  It just can't do 
that without moving first. </description><dc:creator xmlns:dc="http://purl.org/dc/elements/1.1/">Matthew Skala</dc:creator><pubDate>Mon, 22 Aug 2005 01:25:32 -0000</pubDate><guid isPermaLink="false">https://sourceforge.net6d7d42ca4bc1c33a8f27a17464c48b57c22d2a24</guid></item><item><title>Trouble disembarking from transports</title><link>https://sourceforge.net/p/xconq/bugs/119/</link><description>If an occupant of a transport is given an order to "move to" a 
location more than one cell distant, it waits forever instead of 
getting out of the transport and then moving to the location.  
This effect doesn't seem to be universal; in the standard 
game it seems to happen with units exiting troop transports 
and aircraft carriers, but not cities. 

To demonstrate:  Load some infantry into a troop transport, 
put the troop transport in a water cell adjacent to land cells, 
then select an occupant and give it a "move to" command.  If 
the target is one of the adjacent land cells, the occupant will 
correctly disembark; but if it's a cell two or more cells distant, 
it will just sit in the transport indefinitely, never waking up 
again for fresh orders. 

Tested in tkconq under Linux, not sure if the effect is 
interface-specific. </description><dc:creator xmlns:dc="http://purl.org/dc/elements/1.1/">Matthew Skala</dc:creator><pubDate>Mon, 22 Aug 2005 01:17:42 -0000</pubDate><guid isPermaLink="false">https://sourceforge.neta10af4ca38a1689405a696783fb6ba85f2814980</guid></item><item><title>Can't create starving unit</title><link>https://sourceforge.net/p/xconq/bugs/118/</link><description>This is an odd one:  it appears that you can't create a unit \(by 
ACP-independent or ACP-consuming construction, it doesn't 
matter\) if the newly created unit would be in a state of 
starvation.  It is possible that it's actually the "completion" 
action, not "creation", that's the problem.  I noticed it on units 
which ought to be completed in a single action. 

Attached game file demonstrates it.  Attempt to create a 
second army, and your turn ends without a new one being 
created.  Try again, and the game will run out of control.  
Looks like it may be a plan-related issue, since the creating 
unit seems to become "busy" in some way \(thus the turn 
ending\).  If you comment out the last time \(the one that 
causes starvation\), everything works fine.  Another guess 
would be that maybe somehow the "starvation check" code 
is being run before the unit's HP are set, but I haven't been 
able to catch that actually happening in the debugger. 

I know this worked circa March 2005, which is the last time I 
looked at my XcCity game; it seems to have been broken by 
some update between then and now. </description><dc:creator xmlns:dc="http://purl.org/dc/elements/1.1/">Matthew Skala</dc:creator><pubDate>Sat, 06 Aug 2005 00:53:38 -0000</pubDate><guid isPermaLink="false">https://sourceforge.net90e2afd4419360e92f3e773306bd4e6f71c82392</guid></item><item><title>Crash when selecting non-unit</title><link>https://sourceforge.net/p/xconq/bugs/117/</link><description>Start a game \(tested so far in introductory/default game and 
solar.g\) in X11 tkconq \(not tested in other interfaces yet\), 
press "z" to get out of move mode, click on a cell that doesn't 
contain a unit.  Dialog box about values being out of range, 
followed by segfault.  I don't have a lot of time right now - will 
probably do a more detailed investigation, stack trace, etc., 
later. </description><dc:creator xmlns:dc="http://purl.org/dc/elements/1.1/">Matthew Skala</dc:creator><pubDate>Thu, 04 Aug 2005 14:37:12 -0000</pubDate><guid isPermaLink="false">https://sourceforge.net1294d61507e97016c5d4658270cb44ac61a60f0b</guid></item><item><title>Resupplying Units Confused by Two Sources</title><link>https://sourceforge.net/p/xconq/bugs/116/</link><description>It seems that if a unit has two necessary materials
that it can receive from two seperate sources, then it
will go to the closest source of one of those
materials, rather than the closest source of a needed
material.  This can be seen in the attached grue.g
file.  If you play as the Civilized or Dwarf side, you
will have a Dwarven City and a Town.  The Town can
produce Merchants that have 100 Grain and 50 Ore,
whereas the Dwarf City produces Dwarven Merchants that
have 100 Ore and 50 Grain.  Merchants consume 1 Grain
and 1 Ore each turn and starve to death if they reach
0\.  Towns produce Grain and Dwarven Cities produce Ore.  

What I expected to happen would be that the Dwarven
Merchants would travel to the Towns and Merchants would
travel to Dwarven Cities \(And then back and forth as
they ran low\) simulating trade routes.  Instead, the
newly created Merchant and Dwarven Merchant units stay
right next to their creators.  The AI isn't just not
moving them, it states that it is moving them next to
their creators to resupply.

</description><dc:creator xmlns:dc="http://purl.org/dc/elements/1.1/">Elijah Meeks</dc:creator><pubDate>Mon, 01 Aug 2005 17:20:30 -0000</pubDate><guid isPermaLink="false">https://sourceforge.net8338f403f5840e1e013de20c9151c2a986d6cee0</guid></item><item><title>ACP-indep construction can't resume</title><link>https://sourceforge.net/p/xconq/bugs/115/</link><description>I hate to flag yet another ACP-independent construction bug, 
but:  ACP-independent constructors don't seem able to pick 
up work on an incomplete and interrupted construction task.  
Start building something that takes a while, stop in the middle 
to do something else, and then if you issue another "build" 
command, you get a fresh incomplete unit instead of more 
work done on the existing incomplete unit. 

The new solar.g provides a reasonably good test case:  
build an orbital shuttle \(so that you'll be able to step a turn at 
a time even while your initial colony is busy building\), then 
start building a Voyager.  After it's partly built \(should take a 
while if all you've built so far is an orbital shuttle\), give a "build 
a Mariner" command, which will override the "build a 
Voyager" task.  Now you can't go back to working on the 
existing incomplete Voyager; you can only start a fresh one. </description><dc:creator xmlns:dc="http://purl.org/dc/elements/1.1/">Matthew Skala</dc:creator><pubDate>Thu, 28 Jul 2005 04:00:38 -0000</pubDate><guid isPermaLink="false">https://sourceforge.net3904498accbbfdacea4c6eaedbcde8170b21a575</guid></item><item><title>Incomplete units allowed to participate in backdrop economy</title><link>https://sourceforge.net/p/xconq/bugs/114/</link><description>As the title suggests.  Incomplete units seem to become full 
participants in the backdrop economy as soon as they exist, 
at least in model 1, and that's a problem if the unit's main 
function is to act as a relay or storage site - because instead 
of having to pay the full construction cost for it, you can just 
start building it and then have the constructor go do 
something else.  Should be easy to fix, and I plan to; I haven't 
done so yet because I'm not certain what the correct 
behaviour ought to be.  For instance, maybe units should be 
allowed to take but not give material before they're complete, 
so they can start life with normal levels; or maybe \(this is the 
option I'm leaning toward\) they should not participate in 
material exchange at all until fully complete. </description><dc:creator xmlns:dc="http://purl.org/dc/elements/1.1/">Matthew Skala</dc:creator><pubDate>Thu, 28 Jul 2005 03:50:37 -0000</pubDate><guid isPermaLink="false">https://sourceforge.netfae145ad79595ea33c6397402e48fe733b549799</guid></item><item><title>Online help for ZOC tables is backwards</title><link>https://sourceforge.net/p/xconq/bugs/113/</link><description>The patch I just checked in documents the order of indices 
for the mp-to-\{enter,leave,traverse\}-zoc tables, which was 
unclear.  If you want to describe the movement costs 
associated with unit foo moving through the ZOC of unit bar, 
you need to set the entry \(foo bar\), not the entry \(bar foo\), 
even though \(bar foo\) would be the correct order of indices 
for the zoc-range table when setting the size of bar's ZOC as 
applied to foo.  It would arguably be better to use the same 
order of indexes for all four tables, but I was inclined to just 
document it rather than changing it, because some existing 
games depend on the current behaviour. 

However:  the online help indexes the mp-to- tables the same 
way it indexes the zoc-range table, and produces confusing 
text as a result.  A fix could either consist of changing the 
online help code to use the order of indexes used 
everywhere else, or changing the order of indexes \(in the 
kernel, the documentation, and the library games\) to be more 
intuitive and match the online help code.  The latter might be 
nicer, but would be more work. </description><dc:creator xmlns:dc="http://purl.org/dc/elements/1.1/">Matthew Skala</dc:creator><pubDate>Thu, 28 Jul 2005 03:38:21 -0000</pubDate><guid isPermaLink="false">https://sourceforge.net15ccc25154c49d57619f08daac8cb30e8e256eef</guid></item><item><title>Trouble with blocking/locking ZOC</title><link>https://sourceforge.net/p/xconq/bugs/112/</link><description>The tables mp-to-\{enter,leave,traverse\}-zoc are documented as  
taking the value -1 to entirely forbid other units from entering,  
leaving, or traversing the zone of control.  However, that value is  
not consistently respected in the code.  In some places in 
move.c, values  from these tables are run through loops that take 
the simple maximum starting from a minimum of 0, with the result 
that the value -1 is exactly the same as 0.  Using  the value 999 
instead of -1 appears to be an acceptable  workaround, but 
there \*is\* code that explicitly checks for -1 also, so I'm not certain 
999 really has the same effect that -1 was  supposed to.  I'm also 
not sure the handling of ZOCs with radius 0  or 1 is consistent with 
the handling of larger ZOCs; I think the  explicit checks for -1 in 
small ZOCs may hide this bug as long as the ZOCs are small.  
Fixing it in the code is non-trivial because ZOC checking is very 
complicated, so I'm inclined to think that if setting it to \(large\) 
works as a substitute for -1, it may be best to simply document 
that and leave it at that. 

Sample game: my current draft of solar.g \(attached\), where  
black holes have a radius-2 ZOC which forbids exit but has no  
other effect.  It may take a bit of playing to get to the point where 
this shows up \(try see-all to improve your chances of finding a 
black hole\) but you should be able to sail a unit like a sleeper 
ship into the hole's ZOC and not out.  Actually, you can freely sail 
in and out, because the value -1 is treated the same as 0.  </description><dc:creator xmlns:dc="http://purl.org/dc/elements/1.1/">Matthew Skala</dc:creator><pubDate>Tue, 26 Jul 2005 17:59:01 -0000</pubDate><guid isPermaLink="false">https://sourceforge.netabd446ab68ea08fb3308d9eb5b3118400f86e877</guid></item></channel></rss>