[taylor]

I'd commit this already if it didn't modify the pilot files.  I think I should ask permission from the leaders.

No pilot file breakage until after 3.6.8.  And even then it will be the new pilot file code, which is a MAJOR rewrite.  Any pilot file changes made now will not be included in the new code since I can't diff it (the changes are too extensive).  There is a Mantis bug about pilot file issues so just make a patch of what you want to add and attach it, and a small blurb about what it's supposed to do, to that bug and I'll fold it into the new code and commit the whole thing at once when it comes time to do so.

[ARSPR]

Build updated.  The 'next intentions' does something now , but doesn't autoupdate.  That's going to take a lot more work than I expected, but it'll be worth it I think.  Oh, if you don't have anything targeted, it (should) get the attackers of yourself; should I make a separate control for that?

A couple small but frustrating to track down bugs remain.  For one, please test the function of these commands in a nebula with limited awacs.

I'd commit this already if it didn't modify the pilot files.  I think I should ask permission from the leaders.

GREAT. I'm going to download and test it now with the famous "A game of TAG" and the SOC mission where you have to find your mate through nav buoys.

OTOH, I would split it because I, (and I suppose most of the people too), have Auto-targeting always enabled so I nearly never have no active target AND attackers on my own. (In fact I fairly use "Target Nearest Target's Attacker")


EDIT ----------------

Impressions with your new build:

1. GREAT work.

2. One bug. It shows the orders for all targeted ships even enemy ones (within these ones it should be hidden).

3. Other issue. The first time you press it, it keeps your actual target (or no one if you haven't) and only multitargets the attackers. At the second press and next ones, it leaves the current target (which should have happened from the begining) and cycles through the multitargeted group.
(I didn't notice before that your new multitarget commands cycled through the group. I don't know if it is a new feature or I was just asleep )

4. I've tried it within "A Game of TAG" and I haven't found any other crashes or bugs:

4.1. I have left Alpha 2 fighting with the mara wing (which also engages alpha) and I've flown more than 15 Km away (more than radar range for detection of last virgo wing at least). Multi-targeting attackers keeps working   but I can normally target every ship even at this distance, (once you have detected a ship it seems you can always target it even if you are far far away).

4.2. When the last virgo wing arrives, and it isn't shown in radar, it can't target Lucity or any other ship (due an actual FSOpen bug). So I can't check "Multi-target attackers" over this ship. (It just doesn't do anything).

If you have some problematic mission and steps to make it crash or detect bugs, please post and I will try it.


5. Which logic does the code use to select the "attackers"?

I will use an example. I have used FSPort 3.0 "The darkness and the light". I have ordered my wing (3 ships) to attack the Aten cruiser (I don't remember its name, let's call it ATEN). Then ATEN shows 3 red dots in its HUD target box. When my wing reaches ATEN, its cover fighters defend it so you can see how ATEN has only 2 dots, then 1 dot and finally 0 dots because my wing is disengaging ATEN to defend themselves. Through all this time, "Multitargeting ATEN's attackers" always select my full wing. I do feel this is the right behaviour because all of them have the order of attack ATEN even if they're not doing it just at the moment.

But then I order my wing to "Engage enemy", so it means no orders AFAIK. In this situation if I target the ATEN it can have zero, one, two or three dots but the "Multitarget ATEN's attacker's" action doesn't select that exact number of fighters from my wing. Sometimes they're the same but other times they're not.

So, how do you build the group of attackers?



I continue testing... (But not today)

[Backslash]

2. One bug. It shows the orders for all targeted ships even enemy ones (within these ones it should be hidden).

heh, this is semi-intentional for testing.  course now that I've tried it I quite like it (we did have this ability in TIE Fighter), but it's kind of an unfair advantage I suppose.  I'll put it back when done.

3. Other issue. The first time you press it, it keeps your actual target (or no one if you haven't) and only multitargets the attackers. At the second press and next ones, it leaves the current target (which should have happened from the begining) and cycles through the multitargeted group.
(I didn't notice before that your new multitarget commands cycled through the group. I don't know if it is a new feature or I was just asleep )

Yeah, already noticed this and will fix it.  Funny side effect of reusing code from Target Wing... with Target Wing you don't want to change targets the first press
The cycling through the group worked but not always... so it's not exactly 'new' as much as 'improved'.

4.1. I have left Alpha 2 fighting with the mara wing (which also engages alpha) and I've flown more than 15 Km away (more than radar range for detection of last virgo wing at least). Multi-targeting attackers keeps working   but I can normally target every ship even at this distance, (once you have detected a ship it seems you can always target it even if you are far far away).
4.2. When the last virgo wing arrives, and it isn't shown in radar, it can't target Lucity or any other ship (due an actual FSOpen bug). So I can't check "Multi-target attackers" over this ship. (It just doesn't do anything).
If you have some problematic mission and steps to make it crash or detect bugs, please post and I will try it.

Ah!  well maybe that particular bug is fixed.  Thanks.  The other bug... if I knew how to reproduce them it'd be fixed   Will let you know when I have something else for you to test.

5. Which logic does the code use to select the "attackers"?

I'm using some of the same code as the regular 'Target Nearest Attacker', so basically it checks if the ai of (the attacker) has (the targeted ship) targeted.  Problem is, I haven't implemented the idea of 'deselect a target if it's no longer attacking the thing' yet... so maybe that explains some of what you observe?  Try multitargeting the attackers, then selecting something completely different when ordering the 'Engage Enemy', then multitargeting again.

I hope to implement auto-remove-from-list (when they switch targets), and eventually auto-adding new ships that arrive.  Both of those are more complicated -- I'd have to find places from which to call a 'check if this should be added to / removed from a set' function, and without a performance hit of course.  Lots of exploring the code and learning to do.

[ARSPR]

5. Which logic does the code use to select the "attackers"?

I'm using some of the same code as the regular 'Target Nearest Attacker', so basically it checks if the ai of (the attacker) has (the targeted ship) targeted.  Problem is, I haven't implemented the idea of 'deselect a target if it's no longer attacking the thing' yet... so maybe that explains some of what you observe?  Try multitargeting the attackers, then selecting something completely different when ordering the 'Engage Enemy', then multitargeting again.

I hope to implement auto-remove-from-list (when they switch targets), and eventually auto-adding new ships that arrive.  Both of those are more complicated -- I'd have to find places from which to call a 'check if this should be added to / removed from a set' function, and without a performance hit of course.  Lots of exploring the code and learning to do.

Within all my explanation I'm always multitargeting the attackers, (so ATEN isn`t targeted any more), then I manually re-target ATEN again and press multitarget its attackers to rebuild this group of figheters.

I don't know how FS2 selects which the attackers of a ship are, but I can say it's a different way from the logic used to count them and displaying the dots in the HUD target box.

[ARSPR]

More tests with your build:

+ F3 window doesn't work. In this window the Function Key that appears in upper-right corner is erased. So you cannot add/delete Fxx from any ship to make groups. When you press any Fxx key, it briefly blinks in the corner but it's erased.

+ I can't find any logic in how Attackers Group is built. Following with "The Darkness and the Light" mission:

• I order alpha wing to attack ATEN. Then, nearly always, (in previous post I thought it was always), they are selected as ATEN attackers even when they're defending themselves.
  
  When Shivans arrive, (a CAIN and some fighters), they have the order to attack ATEN (your "bug" about enemy orders is really useful, do not turn it off while testing). CAIN seems to be always selected as an ATEN attacker but shivan fighters are nearly never selected as attackers.
  
  During all these tests, ATEN can show any number of dots in HUD target box and they do not have to be the same as the number of attackers. Nevertheless it seems that when CAIN arrives and is always selected as attacker, ATEN has always one dot.
  
  
• In this situation, if I order alpha wing to Engage Enemy (No orders), they are sometimes selected as ATEN attackers. And of course I can't find a relationship between the attackers group and the ATEN HUD dots...

I don't know if you can find any logic through reading the code. AND I don't know if your new code has messed the attacker selection code or probably this was the "official retail" behaviour but it couldn't be detected till now. If this last one is the true one, then you shouldn't change or repair anything about it IMO.

[Backslash]

Good catch with the F3.  Arrgh, I have a general idea what the problem is and it may not be easy to fix.  Or it'll take some creativity.

Updated the build with a Target My Attackers type control, and fixed the Target Attackers not-selecting-the-first-attacker problem.

Hmm, I can't figure out the logic behind how it's choosing the attackers either... all I tell it to do is check if the ai of the attacker has a certain ship targeted.  The AI code is agreed by everyone to be hard to figure out.  I haven't changed the attacked selection code at all.  The way you describe it almost sounds like the AI has more than one target, or can target one ship while doing something else like defending themselves.  Or I wonder if there's a difference between the AI's 'goal' vs their 'target'?  I haven't looked at the dot code yet -- maybe it will shed some light on the situation.

Whoops, my web site is down!  Hmm... ok I'll put it temporarily at http://myhome.spu.edu/calebjackson/FSO/fs2_open_multitarget.exe

[ARSPR]

Well, I'm going on holidays for 2 weeks so I doubt I will be able to test your builds. 

[ARSPR]

Oops a double post.

I've been able to make a fast test (just two missions) and all seems to go fine (but F3 bug) with your new build. (Bye, bye till next week).

(It would be great if these features could reach 3.6.9. launch. But, hey, I really can't and I don't complain about your work in any way. Keep on it  )

[ARSPR]

A triple post, sorry (but a week after the last one, I would update this one during the next days if it keeps in the last position).

I've started testing again your last build and I've noticed the next bugs:

1. The previously known F3 bug. (And the enemy orders one, of course).

2. There's a a cross effect in multitargeting cycling between "Target's attackers" and "Your attackers". For example if you target your attackers then "Target's attackers" also cycles through this just-built group instead of selecting the current target's attackers. The complementary one also happens, so "Your attackers" cycles through the just-built "Target's attackers" group.

3. Strange behaviours with "Your attackers":
3.1. Sometimes "Your attackers" works as expected and it multitargets all your enemies even if they're just one (making even a multitarget group of just one ship).

3.2. Sometimes it just select one of your attackers WITHOUT making a multitarget group (it seems to work just as r key).

3.3. Sometimes the first time you press it, it multitargets all your attackers BUT the closest one, which is only targeted. Then the following pressings cycle through this group but you are missing your closest attacker. The best example of this can be found in FS Port's "The darkness and the light". If you get very close to the Satis transports so they become your attackers. Then the nearest one is not multitargeted. (I allways command my wing to engage the Aten cruiser so they do not interfere with my attackers).
(I will try narrowing this behaviour).


(OTOH, excuse me if I'm really dull about it, but I DO think that the multitarget HUD distance and aiming point additions and the new 'Multitarget target's wing' action would be really nice in 3.6.9. The other two new actions are still too buggy).

[taylor]

(OTOH, excuse me if I'm really dull about it, but I DO think that the multitarget HUD distance and aiming point additions and the new 'Multitarget target's wing' action would be really nice in 3.6.9. The other two new actions are still too buggy).

Not for 3.6.9, it's far too late for that at this point.  If it's not a bug fix then it doesn't get in.  But I'm sure that the unstable branch is hungry for some more code.

[ARSPR]

*bump*

Anything new on this Backslash?

And just another thing. While testing 3.6.9. RC3 with Derelict I think there are strange enemy fighter behaviours. So I think it could be a good idea to add to all SCP releases the "showing enemy orders" feature triggered by a flag in Dev Tool section of Launcher.