Bug: new-style monster list does not correctly report line-of-sight

[jv123]

The new-style monster list doesn't always report line-of-sight correctly. Take a look at this save file. The character has telepathy and so is aware of the pack of hounds in the nearby room, but although he can target 3 of them directly, the monster list reports that he can't see any of them.

Note that /without/ telepathy he likely wouldn't be able to see any, since the room isn't lit --- I guess that's how it's calculating things at the moment. But I think a different algorithm is needed to cope with things like this: for each monster you're aware of, you need to check directly to see if it's in line-of-sight.

Cheers,
JV.

[jv123]

It also reports that you can see a monster, when you can't! Again, an example is attached.

[Magnate]

The new-style monster list doesn't always report line-of-sight correctly. Take a look at this save file. The character has telepathy and so is aware of the pack of hounds in the nearby room, but although he can target 3 of them directly, the monster list reports that he can't see any of them.

Note that /without/ telepathy he likely wouldn't be able to see any, since the room isn't lit --- I guess that's how it's calculating things at the moment. But I think a different algorithm is needed to cope with things like this: for each monster you're aware of, you need to check directly to see if it's in line-of-sight.

It already does that, using player_has_los_bold() - which clearly isn't working. I thought the 'bold' bit meant that it didn't check to see whether the square was lit, so it should have picked up this case.

Thanks for the report - I've opened a ticket for this.

[ajps]

It already does that, using player_has_los_bold() - which clearly isn't working. I thought the 'bold' bit meant that it didn't check to see whether the square was lit, so it should have picked up this case.

Thanks for the report - I've opened a ticket for this.

I think player_has_los() is for actual genuine line-of-sight, not awareness or projectability and all that jazz. If you want everything targetable, use target_able() instead. If you want everything the player is aware of that might be able to get him with a breath, erm... you might have to do that yourself.

[Magnate]

I think player_has_los() is for actual genuine line-of-sight, not awareness or projectability and all that jazz. If you want everything targetable, use target_able() instead. If you want everything the player is aware of that might be able to get him with a breath, erm... you might have to do that yourself.

Hmm. I'll try player_has_los() and see if it picks up this case of monsters sensed by ESP who are not visible to the naked eye. Thanks.

[Magnate]

In conclusion, I don't think there's anything wrong with the monlist code. I think the bug has revealed an underlying problem with the ESP code, in that monsters sensed by ESP do not update cave_info[] as viewable. Will look into this further. I'm working on an assumption that this bug only occurs with ESP - if anyone observes otherwise, let me know.

Fixed in r1439.

[Magnate]

Ok, I have committed r1438 which uses player_can_see_bold rather than player_has_los_bold. I'm not sure if this will work (as it may rely on the grid being lit), but please test with ESP.

[Magnate]

Ok, I have committed r1438 which uses player_can_see_bold rather than player_has_los_bold. I'm not sure if this will work (as it may rely on the grid being lit), but please test with ESP.

That wasn't the answer. The answer is to use (m_ptr->mflag & MFLAG_VIEW), which is designed to record visibility. It wasn't being set properly when monsters in LOS were detected by ESP - fixed in r1439.

[jv123]

Thanks Magnate, a speedy fix as usual!

[Magnate]

Thanks Magnate, a speedy fix as usual!

Well, I gave up and used projectable() in r1441, as I think the LOS concept is broken with respect to ESP.

[PowerDiver]

That wasn't the answer. The answer is to use (m_ptr->mflag & MFLAG_VIEW), which is designed to record visibility. It wasn't being set properly when monsters in LOS were detected by ESP - fixed in r1439.

I wonder if that explains some of the disturb behavior I thought needed a change.

[PaulBlay]

Well, I gave up and used projectable() in r1441, as I think the LOS concept is broken with respect to ESP.

Actually I think projectable may be more useful. Isn't it more useful to know what monsters can hit you with a spell / missile than which monsters you can see?

[jv123]

The original idea was that it should tell you which monsters you can target. That's the most useful thing.

[Magnate]

The original idea was that it should tell you which monsters you can target. That's the most useful thing.

Well, projectable() does that - but Paul should note that it's not symmetrical, so it doesn't necessarily capture everything that can target you.

[PaulBlay]

Well, projectable() does that - but Paul should note that it's not symmetrical, so it doesn't necessarily capture everything that can target you.

if ((projectable(p_ptr->py, p_ptr->px, m_ptr->fy, m_ptr->fx, PROJECT_NONE) || 
        (projectable(m_ptr->fy, m_ptr->fx, p_ptr->py, p_ptr->px, PROJECT_NONE))

OK, now it's symmetrical.