Page 31 of 43

Re: Help needed testing 3.2.x Campaign games!

PostPosted: 11 Oct 2017, 11:17
by -Philosopher-
Bethrezen wrote:ok so just moved the list of general issue find it here

http://forums.wz2100.net/viewtopic.php?f=6&t=13753

Great - thanks. I'll add a post to the thread that references the tickets raised. Suggest we then edit and update those two posts, keeping them up to date with the full list as things are discovered, so they are in effect sticky, with the rest of the thread for discussion on specific issues. Sound good?

@Beserk Cyborg, strictly speaking, a number of these are change requests rather than bugs per se. How should they be handled? Tickets, same as the bugs, or is a different approach preferred?

Re: Help needed testing 3.2.x Campaign games!

PostPosted: 11 Oct 2017, 11:23
by -Philosopher-
Bethrezen wrote:
It's certainly possible. Check the screenshot I attached. I did that on 2.3.8, however. Whether it's still possible or not on later versions (with their changed logic for shooting things on ridges) I'm less sure about. My thinking, however, is it's theoretically possible, so for safety's sake, it probably makes sense to handle this condition explicitly in jscript.


Indeed, it would definitely be a good idea to add a check to address this possibility on the off chance that it does happen because you don't want situations like this braking the game.

question did you actually check to see if your inability to close the drop zone stopped the level from ending because if did then there a new check will need to be added, however if the levels still ends fine despite the inability to close the drop zone then i don't see it being a big problem, it would simply me a minor inconvenience.

Yes, I did, and, weirdly, no it didn't. So, I agree, it could have been worse (all I had to do in practice was deal with a number of additional transports while I finished off prep for subsequent stages). However, I'm thinking better safe than sorry - if it's not too much work to build in the check then maybe it might head off an issue if there are any other stages that require an LZ closed to finish.

Re: Help needed testing 3.2.x Campaign games!

PostPosted: 11 Oct 2017, 12:01
by -Philosopher-
Bethrezen wrote:maybe, mayby not it depends, see when i attack the base from here with my heavy machine gunners.
...
Then the mini pod group doesn't come, also if you take out the scav base with artillery then again they wont come.

However if i attack them from here

Image

Then sometimes they will come as you can see in this screen shot, I'm not sure exactly what triggers them because there trigger seems a little intermittent on v1.10 but it does happen, it might possibly be triggered by me collecting the artefact that's about the only thing i can think of since i had my units set to hold fire so they didn't accidental shoot one of the new paradigm turrets.

OK. So if we are both able to destroy the scav base without angering NP, then clearly that in itself isn't the trigger. It's also not collecting the artefact as I regularly do that without triggering the cutscene. e.g. although they're out of the frame and you can't see it, there are canon armed units in the field when I made my previous screenshots.

As to what else, other than hitting an NP unit or the NP base, is a trigger, it's hard to know. Are you certain you didn't clip the sensor unit or a base wall, even with a single machine gun bullet? One thing I noticed is that even shots that miss their intended targets, but then go on to hit (e.g.) the sensor count, as do shots that are trying to shoot past it - they will hit the sensor if it's in between the shooter and its target. (aside: for this reason return fire doesn't work - even though the sensor has no weapons to fire with, it can still be hit by return fire if it's in the wrong place).

[Edit] Just looked at your screenshot more closely and it's clear this isn't the case in your particular example - not the sensor at least, as it's still far away down south (I presume that's what the lone yellow dot at the bottom of the mini-map is). Thinking about it, I suspect that while getting close to the base doesn't trigger the cutscene or commander, I think it does cause the mini-pod group to come out and investigate. I assume you approached the scav base from the NW in your example? That would be close enough, so that may be it. I'll take another look (I'm curious) but I'm sure @Beserk Cyborg can take a look and form a view anyway (I had a go looking at the scripts, but it's going to be a while before I'm up to speed enough to figure things like this out). I think getting the commander's behaviour right is higher priority anyway.

Re: Help needed testing 3.2.x Campaign games!

PostPosted: 11 Oct 2017, 18:26
by -Philosopher-
Alpha 05 - Mini-Pod Group & Cutscene Triggers

We’ve found a rare difference between 1.10 and the 2.x - 3.1.x wzscript campaigns.

First with 3.1.5...

My starting position for this experiment:
Image
Some things of note:
  • ”You will be punished” cutscene has not yet triggered
  • Cannon units bottom right of frame (so tech has been retrieved)
  • Scav units to the NW have not yet moved (probably not relevant)
  • Both commander and mini-pod groups are in their starting positions
  • Both sensor units are locked onto targets (so opposing units are in range)
  • A single tower of the Scav base remains (the separate one that overlooks the base on the ridge)
I then drove my sensor into the base:
Image Image
Even after getting close enough to the base to be shot at, still the cutscene had not triggered. The mini-pod group (and commander group) only started to act when my sensor unit came within their direct visual range. Shortly afterwards my sensor unit was destroyed (inside the base). Even then the cutscene has not triggered. It clearly needs me to be landing hits on them to do so.

With 1.10…

Attempting to recreate the above scenario turns out to be impossible. While scav movement and behaviour is consistent between 1.10, 2.3.8 and 3.1.5, the mini-pod group isn’t. In 1.10, when the factory containing the cannon tech is destroyed (specifically), the NP mini-pod group exits the base to investigate. It then sits in the position shown in the 3rd screenshot and doesn’t attack further unless you have units in range. This doesn’t occur in 2.3.8 and 3.1.5. For the full situation, it’s best to look at the mini-maps because of the constrained resolution:
Image Image Image Image Image
https://postimg.org/gallery/1bp1s5nlm/
Because they hold position and don’t attack further, it’s entirely possible to destroy the scav base without angering NP in 1.10 too (4th screenshot), but it’s more difficult to retrieve the cannon tech without triggering the cutscene than in later versions because whatever tries to is going to take a beating from the mini-pod units.

Back to 3.1.5, Round 2…

From the same starting position, I destroyed the final tower to eradicate the scavenger base. This triggers the NW scavs to start moving:
Image Image
However, unlike 3.2.x and master, they only move this far then hold pattern (so the NP sensor and the NW scav group share the same destination):
Image
(on 3.2.x and master they attack the LZ instead, which is incorrect)

I managed to destroy them all while still not hitting the sensor. Still no cutscene, and still no mini-pod group. That’s the last of the scavs on this stage. Looks like the difference in behaviour with the mini-pod group was introduced early in the version history.

The NW scav group movement is the same on 1.10 and 2.3.8:
Image Image
They’re more spread out than the weird bunching in 3.1.5 however, so it’s harder to destroy all the Scavs on the stage without upsetting NP (in my 1.10 screenshot above I still haven’t landed a hit on an NP unit, but it feels inevitable under normal conditions, given the inherent inaccuracy of artillery in earlier versions).

With some careful mucking about with a mobile sensor and a few reloads, I managed it, however. This then enabled me to sneak past the mini-pod units and replicate the earlier proximity-to-base test:
Image Image Image
https://postimg.org/gallery/kgqz6xtm/
Again, I was able to drive the sensor into the base, wake up the units inside and have it destroyed by them without triggering the “you will be punished” cutscene, so that’s consistent across all of 1.10 - 3.1.5. It’s quite clearly only actual attacks on NP that elicits the response about attacks (as you’d expect it should).

Personally, while I think 3.1.5 -> 3.2.x/master regressions should be a no-no, I’m more ambivalent about changes from 1.10 introduced earlier in the version history (as is the example of the mini-pods). I guess the purist view is the campaign should reflect 1.10, but I think differences introduced in the conversion to jscam should be the priority (indeed this shouldn’t be a reason, in its own right, for differences at all - that would just seem careless - for this reason, I’m more worried about the differences in Scav behaviour). Suggest we leave it to @Beserk Cyborg to decide how important the 1.10 - 2.x differences are?

That said, I did find the way 1.10 played out marginally more interesting...

Re: Help needed testing 3.2.x Campaign games!

PostPosted: 11 Oct 2017, 19:57
by Bethrezen
you do realise that you can just press T and then type biffer baker and then hit enter to activate the almost indestructible units cheat on v1.10 that's what i do when testing stuff like this because it makes life so much quicker and easier and then i just turn off auto fire and target everything manually, during normal play i don't do this only when testing specific issues.

with regards to specific behavioural differences, I'm not to fussed about that stuff I'm more looking for things that simply aren't working correctly, if you start going into all the minute details of specific behavioural differences then we are going go be here all year, its why i only usually raise behavioural differences when there are substantial and noticeable differences because the level script simply hasn't been coded correctly, alpha 08 is a good example of this as that plays out quiet differently on v1.10 and master less so now with the changes in warzone2100-master-20171003-185446-73e7f25.exe

with regards to the specifics of triggering the FMV by driving in to there base when i did it i did it with my full squad of heavy machine gunners set to hold fire don't know if that would make a difference but again it may well be one of these slightly inconsistent things like with triggering the mini rocket pod group it doesn't always happen because there are some other factors at play.

Having said all this since we aren't the ones actually doing the coding for this I'm fine with leaving Beserk Cyborg to decide how important the behavioural differences are we have provided the information he needs to make a decision so as far as I'm concerned that's job done for this level.

I presume that's what the lone yellow dot at the bottom of the mini-map is

I assume you approached the scav base from the NW in your example?


yes and yes

Re: Help needed testing 3.2.x Campaign games!

PostPosted: 12 Oct 2017, 18:25
by Bethrezen
So continuing my testing of

warzone2100-master-20171003-185446-73e7f25.exe (03-Oct-2017)

Alpha 09

I notice a small glitch on alpha 09 again the enemy units don’t appear to be making use of the repair bay, some of them appear to but others don’t so, it may be that the ones built from the factories do but the ones flown in on the drop ship don’t but obviously it's difficult to tell so which are which so might be an idea to double check that.

Still got the issue with being able to do the cyborg weapons research without first having built the factory but of course we already know about that one but otherwise everything seems fine.

Re: Help needed testing 3.2.x Campaign games!

PostPosted: 12 Oct 2017, 19:41
by Bethrezen
Alpha 10

For the most part this level seems to work ok however as i already noted in my v1.10 play through of alpha 10 the tanks coming in at the southern entry point seem to get stuck on master this doesn't happen on v1.10 so that's something that would be worth looking at, other then that there is just a very slight difference with the transports, but nothing to major you can find the Alpha 10 v1.10 write up here.

http://forums.wz2100.net/viewtopic.php?f=1&t=12714&start=405&sid=1da5f6802584238a4ac36b4954e06960&sid=1da5f6802584238a4ac36b4954e06960#p138464

Alpha 11

Ok so I'm seeing a number of problems with this level mostly pertaining to the behaviour of the convoy which is still misbehaving.

1.) The 4 faster units go speeding off a head of the tracked tank, this is incorrect if you look at the write up i did for alpha 11 on v1.10

http://forums.wz2100.net/viewtopic.php?f=1&t=12714&start=405&sid=1da5f6802584238a4ac36b4954e06960&sid=1da5f6802584238a4ac36b4954e06960#p138509

the convoy is supposed to stay together.

2.) It seems like you have to destroy the whole convoy before the artefact is dropped, somehow the artefact is being magically beamed between member of the convoy this is incorrect when you destroy the artefact carrier, the artefact should be dropped and the other members of the convoy should turn round and attempt to pick the artefact up again.

There are no star trek style transporters in warzone.

3.) The convoy isn't stopping to regroup like it should, again the specifics of this are detailed in the v1.10 write up of Alpha 11 i did previously.

4.) if you are emulating the v1.10 victory conditions then the level should only end when you return to LZ.

So all in all i think alpha 11 needs another pass.

Re: Help needed testing 3.2.x Campaign games!

PostPosted: 13 Oct 2017, 14:47
by Bethrezen
Alpha 12

Ok so there is still something not working correctly with this level because yet again I have been stymied by a massive build up at there base, which simply shouldn't be happening, and it not like i didn't leave a squad as land based units.

Image

Which means that yet again I'm unable to complete this level, because it simply takes to long to try and whittle them down, and i run out of time, so at the moment the only way i can complete this level is to cheat.

What is even stranger is that when i reload the save i made just before i took the previous screen shot i end up with there entire army coming at me at once and i get crushed, as you can see in this screen shot.

Image

now obviously in this screen shot i have the biffer baker cheat on because its the only way my units can survive against that.

I've attached the save where this happens.

Alpha 12.7z
(91.54 KiB) Downloaded 11 times


GRRRR......

Man I'm really really starting to hate this level, because i don't have any where near this amount of difficulty completing this on v1.10.

Re: Help needed testing 3.2.x Campaign games!

PostPosted: 13 Oct 2017, 17:45
by Berserk Cyborg
Fixed factory/transporter reinforcement group repair. It might make the few missions that use repair facilities a bit tougher. Alpha 5 commander has a much larger defense zone. Alpha 11 forces return to lz and now ends when enemy transport leaves with artifact. I will mess with groups going back to the artifact later.

Think you could use the deity cheat on Alpha 12 and watch for the groups stopping for whatever reason?
Edit: Deleted broken mod.

Re: Help needed testing 3.2.x Campaign games!

PostPosted: 13 Oct 2017, 18:43
by -Philosopher-
Bethrezen wrote:1.) The 4 faster units go speeding off a head of the tracked tank, this is incorrect if you look at the write up i did for alpha 11 on v1.10

http://forums.wz2100.net/viewtopic.php?f=1&t=12714&start=405&sid=1da5f6802584238a4ac36b4954e06960#p138509

the convoy is supposed to stay together.

This is true only because formation speed limiting (F11) exists in 1.10 - i.e. it's not specific to the convoy (as you suspected in your write-up). The difficulty I see here is to restore this without imbalancing the stage, it needs to be restored as it was in 1.10 - i.e. globally, and so groups move smoothly at a continuous speed. It doesn't work to put it back piecemeal. When it applies globally, there's a good chance the player units will be slowed to some degree too. If NP are the only ones slowed, it makes them too easy to intercept (i.e. far easier than it ever had been in previous versions). As this is the primary objective of this stage, that would seem a bad thing.

This might come as a surprise coming from me, but I think this is best left alone at the moment. Other factors - balance (stats/damage/armour) differences and ability to shoot things on ledges - makes a far greater difference to the experience of this stage, so to truly put this back to the 1.10 experience a lot more, in addition, would have to change anyway.

Question to @Beserk Cyborg -

While I appreciate "Whenever different versions of wzcam are found to behave differently (say, 3.x is found to be different from 1.x), jscam attempts to mimic the older version." (from the ToR - http://developer.wz2100.net/wiki/jscam), to restore this properly seems like a lot of work (in this case) and regressions introduced by the conversion to jscript should perhaps have a higher priority anyway?

I might be reading too much into it, but this viewtopic.php?f=35&t=11872&start=165#p135786 seems to imply there's more concern about the direct port than returning to the 1.10 state of the campaign as well. Is that correct (or does this not represent the current thinking any more)?

Re: Help needed testing 3.2.x Campaign games!

PostPosted: 13 Oct 2017, 19:39
by Bethrezen
This is true only because formation speed limiting (F11) exists in 1.10 - i.e. it's not specific to the convoy (as you suspected in your write-up). The difficulty I see here is to restore this without imbalancing the stage, it needs to be restored as it was in 1.10 - i.e. globally, and so groups move smoothly at a continuous speed. It doesn't work to put it back piecemeal. When it applies globally, there's a good chance the player units will be slowed to some degree too. If NP are the only ones slowed, it makes them too easy to intercept (i.e. far easier than it ever had been in previous versions). As this is the primary objective of this stage, that would seem a bad thing.

This might come as a surprise coming from me, but I think this is best left alone at the moment. Other factors - balance (stats/damage/armour) differences and ability to shoot things on ledges - makes a far greater difference to the experience of this stage, so to truly put this back to the 1.10 experience a lot more, in addition, would have to change anyway.


True but you forget that the player can turn off the formation speed limiting, which to the best of my knowledge affects only the players units, the other issue with your assumption here is that you are assuming the player is using units with different types of propulsion which in my case simply isn’t true all my units use half tracks on this level so whither formation speed limiting is on or off makes absolutely no difference to me as all my units move at the same pace any way.

However I take your point and I do realise that this is a tricky one to get right given the removal of the formation speed limiting however there are other options here.

If the biggest hurdle here is getting all members of the convoy moving as a group and moving smoothly at the same pace then the answer is simple give them all the same propulsion, because under the circumstances that is going to be the most straightforward work around.

I know from the purist point of view this may not be the most desirable option but realistically I don’t see a better way to handle this that wouldn’t involve masses and masses of work to fix.

Simply changing the propulsion on that 1 tracked unit to half tracks would largely solve the issue and allow you to set the level up so that it plays out like it's supposed to even given the removal of the formation speed limiting, and then on insane you could slightly reduce the time at which the convoy starts moving just to make life a little bit tougher for the player because right now having the convoy not start moving for the first 2:00 to 2:30 minutes is way to easy.

With regards to balancing, if you wanted to preserve that the best way to do that is to make the 2 lancers slightly tougher by giving them half tracks instead of wheels, and that will off set the fact that the tracked tank has become a bit weaker by being given half tracks.

you have to admit of all the possible way to handle this, this is the one that makes the most sense and the one that would be the most straightforward to implement and require the least amount of work.

Think you could use the deity cheat on Alpha 12 and watch for the groups stopping for whatever reason?


I've actually tried this before without much luck but i can certainly try again.

Re: Help needed testing 3.2.x Campaign games!

PostPosted: 13 Oct 2017, 20:30
by -Philosopher-
Berserk Cyborg wrote:Fixed factory/transporter reinforcement group repair. It might make the few missions that use repair facilities a bit tougher. Alpha 5 commander has a much larger defense zone. Alpha 11 forces return to lz and now ends when enemy transport leaves with artifact. I will mess with groups going back to the artifact later.

I took a quick look at Alpha 5 with the latest mod. It's still not right unfortunately:

  • The commander is supposed to attack if his base is attacked, but not before. Here, he's (incorrectly) springing into action when an NP unit (only) has been attacked:
    Image
    (this triggers the cutscene and mini-pod group - if you don't want to revert it to 1.10 behaviour - but shouldn't wake the commander yet)
  • Here, the commander is (incorrectly) springing into action merely because I am too close to the base. This shouldn't actually trigger anything:
    Image
    (no hits have been landed on any structures yet)
  • However, let's briefly pretend he is activating correctly... Is the area he's defending still a little small? I'm pretty sure my units would still be being attacked if they were this far north previously, but here's he's gone back inside the base:
    Image
    How does the area compare to what it was in wzcam? I'm also wondering if it's not about area in previous versions and more about traditional 'guard' behaviour - i.e. you have to get out of his range for him to break off the attack, otherwise he'll keep attacking regardless of where you are. Or perhaps a combination of both... not sure... In any event, he's still giving up too easily at the moment (although at least he makes it out of the base now).
  • What level are units attached to the commander (and additional units produced in the factory) set to retreat at? It could be I'm just too efficient at wiping them out, but they seem to be waiting too long to retreat. It was much harder to strip the commander of his associates in earlier versions.
  • Is there a reason the commander wants to close to point-blank range with whatever he's attacking? That just gives him (and his associates) further to retreat when they need to:
    Image
    Here the commander seems to be trying to force his way to the LZ, except I assume he won't go the whole way and will turn around again once my units leave his zone?
  • Units from the factory seem to wait and only attack when the commander does. In previous versions, they'd sally forth and attack on their own once a group was large enough (after NP are activated with the cutscene). The factory produces a lot more units than it did on previous versions too. Between these two differences, you end up fighting them all at once when trying to assault the base, whereas in previous versions if you were clever about it you could handle them in separate groups.
  • The mini-pod group is attacking the LZ here. It should be going after my (attacking) units:
    Image
  • Did you change anything about the way the Scavs behave on Alpha 5? They're still behaving weirdly, but I'm wondering if it's different weird now. Here the scav base periodically sends out small groups to attack my LZ (they didn't use to do that... although what they should actually be doing is accumulating and waiting to counter-attack - along with the scavs that currently just and wait to be bombed on the hill - whatever attacks their base):
    Image

Re: Help needed testing 3.2.x Campaign games!

PostPosted: 13 Oct 2017, 20:47
by Bethrezen
OK so assuming I'm not doing something wrong that mod seems to be buggy

copied the mod to c:\Warzone 2100_portable-master\mods\campaign

created a short cut with the following command line
"c:\Warzone 2100_portable-master\warzone2100_portable.exe" --mod_ca=Updated-Campaign.wz

Loaded up my alpha 10 end of level save finished the level and then went on to Alpha 11 and I can't end the level, wiped out everything on the map and collected the artefact but get no return to LZ message.

i do get errors though here is the log

Re: Help needed testing 3.2.x Campaign games!

PostPosted: 13 Oct 2017, 20:54
by -Philosopher-
Bethrezen wrote:If the biggest hurdle here is getting all members of the convoy moving as a group and moving smoothly at the same pace then the answer is simple give them all the same propulsion, because under the circumstances that is going to be the most straightforward work around.

Why is it so important for them to move as a group at all? It doesn't affect the overall experience of the stage - which is fairly consistent across all four versions of the game I've been looking at (more than I can say for many of the other stages we've been testing) - possibly because of the way the stage is designed, the artefact group is back together in the place they're most likely to encounter the player units anyway. Indeed, past attempts to make them move as a group made this worse rather than better.

I strongly disagree with messing with the propulsion - that changes all sorts of other things (unit strength, stage timing, etc.), and if you start messing with even more things to compensate you just end up with more and more unintended consequences.

I think it's best just left alone and maybe picked up if/when formation movement gets looked at as a whole. There are more impactful differences elsewhere that should probably take priority anyway.

Re: Help needed testing 3.2.x Campaign games!

PostPosted: 13 Oct 2017, 21:03
by Berserk Cyborg
Oops, here is the corrected version.
Updated-Campaign.wz
(98.43 KiB) Downloaded 12 times


-Philosopher- wrote:While I appreciate "Whenever different versions of wzcam are found to behave differently (say, 3.x is found to be different from 1.x), jscam attempts to mimic the older version." (from the ToR - http://developer.wz2100.net/wiki/jscam), to restore this properly seems like a lot of work (in this case) and regressions introduced by the conversion to jscript should perhaps have a higher priority anyway?

I might be reading too much into it, but this viewtopic.php?f=35&t=11872&start=165#p135786 seems to imply there's more concern about the direct port than returning to the 1.10 state of the campaign as well. Is that correct (or does this not represent the current thinking any more)?
At the risk of putting words in vexed's mouth, I think, that for now, the main concern is that the missions work and do not throw error in the logs and missions play similar as before.