Minefields and Rub1.44 (Cont'd)

[Stiebler]

This is a continuation of the thread entitled "RUb1.44 and Minefields" which began in the main forum and is now closed to new entries. By the end, it had become an analysis of Crash-to-Desktop when patrolling off Hull (which never happens in the original SH3 with 1.4 patch). It seems more appropriate to continue in this forum. New viewers should consult original entries in the main forum.

I've been carrying out more patrols with RUb 1.44 in July 1940 on a line 100 km out from Hartlepool and Hull, ie, outside minefield range. What I have discovered is that the CTDs are *much* more common if GB minefield#8 has its density set to 100, than when it is set to 1. In the latter case, it is very hard to provoke a CTD, and all the examples I encountered involved preliminary sighting of a warship first. But one can sail along the 100 km line at a time setting of 512 for weeks of game time before anything happens. And then it cannot be reproduced reliably.

When the minefield density is set to 100, CTDs become far more frequent and finally even *reproducible*! I have a saved position when the crash will occur within one minute of loading it and following the pre-plotted path. But this time the crash does *not* involve preliminary sighting of a warship (although a warship can be heard moving away if I dive to 25 metres). Moreover the crash can be reproduced whether the U-boat is on the surface or submerged.

And, most conclusively of all, the crash does *not* occur if the saved position is reloaded with the minefield#8 density reset to 1.

Obviously there must be interference caused in some way by the minefields to the running of the game. The greater the minefield density, the greater the interference.
1. This might be due to a program bug, because it cannot handle minefields properly (I mentioned in an earlier post that the original SH3 *never* sets the radius of a minefield, whereas RUb 1.44 is full of minefields with radius set to varying values).
2. The problem might be due to Allied ships hitting their own minefields, which the program doesn't know how to handle.

How can Allied ships hit their own minefields?
RUb 1.44 has made the hulls of some ships deeper, and in gale force storms the sea moves by up to 3 metres. This might just be enough to bring the ships randomly into contact with minefields laid at 15 metres depth - and would explain the greater number of CTDs in the region of minefield #8 when the density is set to 100 - in fact, that might well mean several simultaneous hits on one ship, which might also cause instability.

How come so many CTDs appear to be irreproducible?
Maybe the merchant ships randomly spawned around the U-boat are also blundering into minefields, but especially high-density minefield#8.

How come my latest CTD *is* reproducible?
Some warship or other is running through the minefield and keeps triggering the mines when at high density, but not when at low density. I'd guess that the original programmers implemented minefields by a random number generator - when in the field, if the random number multiplied by the density factor exceeds a trigger value, bang! Most of the time, at low density, the warship gets away with it. Rarely, it doesn't.

Why do the crashes occur under different conditions (but especially when using UZO)?
Underlying program bug, probably causing a variable overflow which somehow affects the view. It did occur to me that the periodic crashes with the UZO might be due to something else entirely, such as a defective skin for the ship being sighted. I believe too that I encounter more warships when minefield#8 is set at high density than at low density, and again it might be that a minefield bug is somehow spawning a rogue and unstable destroyer.

What if my theorizing is right?
There will always be a few scattered random crashes when the U-boat is near any substantial minefield. The lower the mine density, the less likely they are to occur.

What if I'm wrong?
SEP (Someone Else's Problem - I can't think of anything else).

Solutions?
1. As a practical measure, lower all minefield densities at once, and especially that of minefield#8! This is done most safely in the Mission Editor.
2. It might be a good idea to reduce hull depths of all ships to original values.
3. Perhaps check the modified skins/shapes of all modified warships.

Other items:
Beery has pointed out that some of the waypoints of minefields don't seem to be connected, and that some of the minefields have duplicate names (including the infamous #8). These may also need attention.

Stiebler.

[oRGy]

Good work Stiebler. Remind me to hire you as a beta-tester sometime. ;)

[Kpt. Lehmann]

If we choose not to use the mission editor... which file needs to be modified to reduce minefield density?

[Sailor Steve]

Unfortunately the Mission Editor seems to be the only way. I reduced British Minefield #8's density to 1 using the SCR file directly, but when I went to the Editor it still said 100, so I had to do it there too.

#8 was the only one I found that was at 100, but then I might not be as thorough as some folks.

[Kpt. Lehmann]

Hmmm, this might be a dumb question Sailor Steve, but did you save your change before closing that file?

I've made myself feel stupid more than once by forgetting to do just that.

[Der Teddy Bar]

Stiebler,
Good work.

I had redone the entire East Coast, the minefields are as extensive as they were during the war. I even have the deeper gaps, which are still mined.

Where as Sailor Steve had only the outlines and a small presence at Dover, I have the entire area filled and Dover. All as per the war map I have. The East Coast minefields are about 35+ k's wide with a density of 10 so as to allow a reasable, but very low chance for the u-boat to cross the entire width submerged.

I had set the depth of the minefields so as to be deep enough for an escort but not a ship. The idea, as it was in the war, was for the shipping to move inside the minefield barrier.

It was, and may still be, if there was actually an interest, to do the mines as historically correct as possible, even to the minimum depth.

For example, I have even included some historical correct Dutch minefields in Holland's Territorial waters for 1939-1940.

PM me and I can e-mail you the file to test. I am certain that it will not crash.

[Stiebler]

Thanks for the offer, Teddy Bar, and congratulations on your labours. I've sent a PM.

Belated thought:
I'm still concerned about the possibility that random spawning of Allied ships with RUb-modified deeper hull-depths might result in these ships hitting minefields with uncertain effects on the original SH3 code - perhaps triggering random crashes.

Rather than modify the hull-depths again, it would probably be simpler, more efficient and more elegant to lower the minimum minefield depths by 5 metres. For example, a minefield originally set from -15 to -25 metres would be changed to -20 to -25 metres.

Now what happens, I wonder, if a minefield is set with depths of -15 to -25 metres in a sloping coastline, where the sea depth suddenly shallows to -20 metres? Probably nothing; doubtless the code only tests whether the U-boat's hull lies within its range without testing the sea depth.

[Sailor Steve]

Hmmm, this might be a dumb question Sailor Steve, but did you save your change before closing that file?

I've made myself feel stupid more than once by forgetting to do just that.
Hmmmmmm.......could be :88) I'll never know now.

@ Teddy Bar: I did the best I could with the maps I had. My map of the Channel only showed Dover as having two single lines across it. Also, I had no clue how to make the minefields 'solid'; the only thing I could get it to do was use the waypoints to make the 'hollow' versions you see. As I said long ago, If you have the info and the means to do a better job, I'm all for it! I just hope we can convince the Ops-Mod guys to incorporate your version once it's done.

As to the density, I think 1 looks pretty good for the big fields. Just my opinion, though.

[Beery]

Hmmm, this might be a dumb question Sailor Steve, but did you save your change before closing that file?

I've made myself feel stupid more than once by forgetting to do just that.
Hmmmmmm.......could be :88) I'll never know now...

I changed the density in the text editor for RUb 1.45, and I just checked - it worked fine.

[Beery]

...Now what happens, I wonder, if a minefield is set with depths of -15 to -25 metres in a sloping coastline, where the sea depth suddenly shallows to -20 metres?...

What if this is the problem? What if, when the mines enter the U-boat's spawn range, they spawn and some mines spawn in the sea bed, thus causing a crash?

[Beery]

Plus, isn't minefield 8 an especially big minefield? What if its very size is the problem - all of a sudden, when the minefield meets the spawn radius, the player's computer has to place hundreds of mines, thus increasing the chance of a lockup or crash?

[Stiebler]

Teddy Bar kindly provided a copy of his new list of minefields (incorporating old and new work) for inclusion in the campaign_SCR.mis file as a replacement for the older minefields.

He'd checked that the new minefields were seaward of the merchant traffic areas (as in real life). Also the minefields have been checked to ensure that they don't collide with the sea-bed. (Was the campaign_RND.mis file checked as well?)

Tested as below:

32 game days per test (July 20th - August 20th 1940) on patrols 100-120 km off the coastline of east Britain between Hartlepool and Hull (or rather, a little further out; the new minefields extend further east than those in RUb 1.44, and I got caught in them once).

For comparison:
1. Unmodified SH3 with patch 1.4: no CTD ever seen (tests and normal game play).
2. Standard RUb 1.44: frequent CTDs, all conditions, some reproducible, one saved as "testpos".
3. RUb 1.44 with GB minefield density#8 set to 1: 1-2 CTDs per calendar month, none immediately after "testpos".

New tests:
4. RUb 1.44 with old minefields totally removed in campaign_scr.mis and the new minefields installed: one CTD after 28 days. No CTD after use of "testpos".
5. RUb 1.44 with no minefields at all in campaign_scr.mis: no CTDs seen. No CTD after use of "testpos".
6. New minefields only (nothing else) in campaign_scr.mis: no CTDs seen. Can't install "stdpos".

Comments:
These data are a bit difficult to interpret. A 32-day patrol is completely arbitrary, and if I'd picked 28 days I'd probably have missed the CTD in case (4).
Ignoring the original RUb problem, all CTDs involve a visual sighting (usually by UZO) of a warship. Does this mean there is a separate bug, unrelated to the minefields? Probably, but it ought to occur in wider play and I haven't used RUb 1.44 enough in normal game play to be sure (I've never yet seen it outside the Hull patrol area). And it is hard to check a visual bug when the visibility keeps randomly changing.
Are the new minefields completely bug-free? Probably (?), but if the crash is precipitated by visual sightings, then there is a problem with a campaign_SCR.mis file containing only minefields: you don't get to see any warships! I encountered precisely one randomly-spawned merchant ship and one aircraft during the whole of patrol (6).

The new minefields:
The new minefields are exceedingly good - I had a look at them in the mission editor before doing anything else - and almost a work of art as well as being evidently the result of enormous labour. Since they also add to RUb's realism, I'd be very reluctant to discard them, especially since whether they really are causing the UZO-CTD is still an open question. And even if there is cause and effect between minefields and UZO-bug, it is now certainly very rare. I'll continue RUB 1.44 play with these new minefields in place.

I do think that it is necessary to provide some kind of warning of the position and size of the *defensive* minefields. There are periodic vague reports of "crashes" in RUb 1.44, and I'd bet that a lot are explained by the sudden unexplained damage, flooding and almost instantaneous sinking that occurs at high time compression after a mine hit. Both Germans and British published details of their defensive minefields, so as to avoid the ill-will that would undoubtedly occur if a neutral ship were to sink on them. So this is perfectly realistic. The minefield designer is obviously the right person to provide the description (textual would doubtless suffice).


Beery asked, in effect, whether mines are being spawned and the consequences if laid in waters of varying depth.

Certainly this might cause serious complications. But surely the programmers would not keep checking the position of literally every mine in a minefield? That would be incredibly inefficient when a random number generator would suffice (random number x minefield density - who could tell the difference between this result and that of a precisely placed unseen mine?), and it would hugely slow the game in the vicinity of a major minefield.

It's a pity we can't get clarification of basic issues like this (or whether there are any known issues with minefields) directly from the Romanians. Does anyone have a direct contact?

Stiebler.

[Beery]

Can someone put together a new SCR layer based on the RUb1.44 SCR layer but with these new minefields added and the old ones removed? I'd do it myself but I have no idea regarding the SCR layer and I'd be afraid of making a mistake.

[Kpt. Lehmann]

Can someone put together a new SCR layer based on the RUb1.44 SCR layer but with these new minefields added and the old ones removed? I'd do it myself but I have no idea regarding the SCR layer and I'd be afraid of making a mistake.

Yes please do... and make it available individually as well?

[Stiebler]

Beery wrote:

Can someone put together a new SCR layer based on the RUb1.44 SCR layer but with these new minefields added and the old ones removed? I'd do it myself but I have no idea regarding the SCR layer and I'd be afraid of making a mistake.

You can have my current copy.

But Der Teddy Bar (originator of the new minefields) must give his permission first. Teddy Bar - please send me a PM of confirmation.

[Beery]

But Der Teddy Bar (originator of the new minefields) must give his permission first...

Agreed.

[Stiebler]

A further very necessary test patrol:

7. Added Teddy Bar's new minefields to the original SH3 with patch 1.4 (after removing original minefields) in file campaign_scr.mis. Patrolled between Hartlepool and Hull in usual way until fuel ran out (41 days). No CTDs at all, despite plenty of traffic of all sorts.

There must still remain some bug in RUb 1.44 which causes the UZO-CTD when certain warships are sighted.

Possible candidates:
a) deeper hull depths of some ships, striking minefields and causing strange results elsewhere, as previously speculated.
b) defects in ship model/skins for some particular class of warship.
c) overcrowded escort groups. (Noticeable with created missions that ships too close together tend to run amok, in all directions, instead of following plotted route.)
d) defective set-up for a single destroyer or other warship. I've checked campaign_scr.mis in the Mission Editor for warships crossing the Hartlepool/Hull line, and can't see any obvious faults.
e) Anyone got any other ideas?

Stiebler.

[Der Teddy Bar]

For all of my minefields I have checked the depth to ensure that none are too deep. I have checked the density for both framerate impact and potential crashes.

The density in the end was decided upon by testing and arriving at a figure that I thought would be reasonably accurate without being an uber weapon.

I have set the minimum depth, the dates of exisitance and width to the historical if known. If the historical depth exceeded the depth of the water, I adjusted the depth to suit. If the historical depth was too shallow for escorts it was adjusted it to suit.

Stiebler is mistaken in that I "the new minefields were seaward of the merchant traffic areas". My first and only priority was historical accuracy.

I had hoped that the project would have had some interest, but it did not. So I stopped work on it as it would require that the merchant ships paths be modified to be within the East coast minefields along with scripted escorts and aircraft to make the minefield effective.

Also, the time taken to attain historical information is immense. There is little to no information available.

Beery now has the work in progress minefield and will also get any updates that I produce for use in RUB.

It should be noted that I did ensure that the depth of the shallow minefields was sufficient to allow all the escorts free travel through them.

Thankyou to Stiebler and Kpt. Lehmann for taking the time to test my minefields, and for your input.

[Beery]

I had hoped that the project would have had some interest, but it did not...

I think the lack of interest in your work that was apparent in the SH3 Mod Team forums was due to the fact that the Ops mod team - the guys who were working on the SCR layer - had pretty much split up (or at least I believe they have - I haven't heard anything from them for a while). I was waiting for one of them to say something because I thought they were still working together, but it seems they weren't.

Hehe, now they'll probably all prove me a liar by resurfacing and claiming they've been in their underground lair plotting version 2.0. :-j