SH3 Bug caused by over-long RWR ranges.

Stiebler · 07-15-09, 10:59 AM

U-boats sailing out of Bergen in late 1944 or 1945, in the NYGM mod, have been increasingly afflicted over several releases by sudden air crashes. The symptom is always the same: the U-boat is fitted with Tunis, is sailing on the surface, is attacked by a long-range bomber near the Faeroe Islands (where there is an airbase), and the U-boat gives a warning of radar detection. Almost immediately afterwards, the aircraft vanishes and is replaced by a 'ship-sunk' symbol on the map. A cruise from Bergen to square BD can sometimes result in a trail of 'sunk ships' along the U-boat's route!

The problem is intermittent and random, and originally so very rare that it escaped detection during testing, and subsequently was sufficiently unusual that it seemed to be a program bug. However, as NYGM became more complex, the problem became more noticeable. It was when air crashes occurred in late 1943, with bombers based on the airbase in the Azores crashing after radar detection with Metox, that it seemed necessary to investigate further.

It took ages to discover the cause of the problem, involving changing airbases to display single types of aircraft and slowly adding new features back into a stock version of SH3. At first, it seemed to be a cloning problem for different aircraft/radar/RWRs, or a problem with bomber load-outs. However, to cut a long detective story short:

The SH3 program has an intermittent problem with radar warning ranges set to 30,000 metres or more. For NYGM, this affects the Metox and Tunis radar-warning receivers (RWRs), which both have a range of 30,000m. Actually, in real life, both had a range of around 50,000 metres, but the range was cut down in NYGM to 30,000 metres, which is the limit of the real-life horizon.

The extraordinary thing about this program bug is that (with NYGM), it only affects the standard Liberator. The Sunderland and Halifax, both long-range bombers with the same radar equipment in November 1944, are unaffected. (Typical of a possible memory leak in the code, devs please note.)

If the ranges for Metox and Tunis are set to 24,000m, the air crash problem disappears completely. After extensive testing, it is clear that this range is easily sufficient for the U-boat to dive to avoid an aircraft, even at time-compression = 1024.

I intend to release a patch, with other small fixes, for NYGM shortly. Anyone who has used in the past the ancient (2005) Stiebler 'Radar_Detection' fix (a stand-alone mod, also used in the old RUb mod from which other super-mods have been derived subsequently) should use a hex editor (or, better, SH3 Mini-Tweaker or SH3 File Analyzer, or S3DEditor) to change both MaximumRange and PreciseRange for the Metox and Tunis RWRs to 24000 from the existing 30000 (thus four changes in all).

The affected file is silenthunteriii\data\library\sensors.dat.

I have also checked GWX3 RWR ranges. These are set for Metox to 22000 and for Tunis to 25000, so probably users of GWX do not need to make any changes.

Stiebler.

coronas · 07-15-09, 12:44 PM

Thanks for the tip, Stiebler.

NGT · 07-15-09, 03:43 PM

Sir,

I meet hard difficulties to write in English, and my presence in the forum is limited due to this problem.

But, I seize this opportunity to express my warmest thanks for your global contribution for the improvement of playability and diversity of this game.

Your reports were always clear, well draw up, analytical as much are necessary, interesting and useful TO EVERYBODY.

You are not limited to improve NYGM, but you share the fruit of your research with the whole community, and you offer your knowledge for improving the technical level of each one from us, and, by the same way, for improving the entertainment we need after a day in real life’s problems.

All that, without any behavioral conflict, without any offense.

Your attitude is an example to follow, here and in real life. Your stance go beyond NYGM (a very serious compilation indeed), beyond SH3.

Certainly, a real gentleman.

onelifecrisis · 07-15-09, 11:10 PM

Yeah +1 to what NGT said (minus all the flowery stuff

).

Kpt. Lehmann · 07-16-09, 01:10 AM

I concur with the findings posted on this thread... and can now confirm (although to a slightly lesser degree) Stiebler's assessments, though it has been a rather long time since I've observed the situation referred to regarding inexplicably downed aircraft.

(As I recall : )

When the RWR ranges were set farther away to match with historical values, I witnessed a handful of oddities that in retrospect may have been related to overlong RWR ranges set in relation to SH3's limitations. However, they were infrequent enough... that I simply didn't recognize them as a consistent problem... and/or thought maybe it was just my own machine/installation. It simply wasn't 'clear' to me at any rate.

Rubini attempted to persuade me to lower them for other reasons which I understand better now... and which now compliment this situation as a possible fix.

Currently I use his "Unofficial RWR Range" fix which reduces RWR ranges in GWX further than those accurately listed by Stiebler above... and is accurately described by Rubini as (paraphrasing) 'Making it no longer possible to have a cup of coffee before diving' when under attack by air.

If I can take a bit of liberty here without naming sources... (please feel free to refute... its late and I should be in bed at the time I am writing this.) ... It is my feeling/understanding based on anecdotal accounts of U-boats under air attack... that RWR units were far from 100% effective... and that ranges given are likely a total maximum that could be expected in perfect conditions. In my experience with Rubini's RWR mod, it seems to me that recent events during my patrols... now more closely match those anecdotal accounts.

Secondarily, I have found that the associated modifications (optional inclusions IIRC) to AI rocket-firing solutions work as advertaised...

...AND something that I'm not sure even Rubini is aware of...

Using the same mod, units using aircraft bomb and DC's have fewer instances of "failures to drop ordnance" on first passes unlike before. Aircraft will still make multiple passes... but not quite as many... given that ordnance is dropped sooner and generally close enough to be a threat to the boat. (Either causing actual damage or at least rocking the boat via near miss.) It seems as if the ballistic solutions plotted by the game are now a bit more 'loosened-up.' This condition should also result in fewer aircraft being shot down by U-boats in-game by those who DO choose to engage... and thereby further encouraging players to dive as opposed to engage... given that the consequences are a bit more costly. (Fewer repeating passes by aircraft and greater chance of damage by ordnance dropped during first pass.)

It wasn't my intention to derail the thread by going OT with rockets and bombs... but I think that little package of mods built by Rubini is bearing out to be something we should maybe consider/request to be "official" augmentations to the GWX and NYGM mods respectively.

irish1958 · 07-16-09, 07:13 AM

Kpt. L, I fully agree; Rubini's mod is terrific and the pyrotechnics are spectacular.

Stiebler · 07-17-09, 11:48 AM

The patch that I promised above for NYGM, together with other minor fixes, is now available as NYGM 3.3 from my signature below.
Or see:
http://www.subsim.com/radioroom/show...&postcount=247

Interesting observations from Kpt Lehmann. One can mimick unreliability in the RWR by providing a longer 'sweep period' in sensors.dat for each item, Metox, Borkum, Naxos and Tunis.

Thanks for the kind words, NGT and OLC.

Stiebler.

07-15-09, 10:59 AM	#1
Stiebler Fuel Supplier Join Date: Oct 2005 Location: London, UK Posts: 1,237 Downloads: 29 Uploads: 4	SH3 Bug caused by over-long RWR ranges. U-boats sailing out of Bergen in late 1944 or 1945, in the NYGM mod, have been increasingly afflicted over several releases by sudden air crashes. The symptom is always the same: the U-boat is fitted with Tunis, is sailing on the surface, is attacked by a long-range bomber near the Faeroe Islands (where there is an airbase), and the U-boat gives a warning of radar detection. Almost immediately afterwards, the aircraft vanishes and is replaced by a 'ship-sunk' symbol on the map. A cruise from Bergen to square BD can sometimes result in a trail of 'sunk ships' along the U-boat's route! The problem is intermittent and random, and originally so very rare that it escaped detection during testing, and subsequently was sufficiently unusual that it seemed to be a program bug. However, as NYGM became more complex, the problem became more noticeable. It was when air crashes occurred in late 1943, with bombers based on the airbase in the Azores crashing after radar detection with Metox, that it seemed necessary to investigate further. It took ages to discover the cause of the problem, involving changing airbases to display single types of aircraft and slowly adding new features back into a stock version of SH3. At first, it seemed to be a cloning problem for different aircraft/radar/RWRs, or a problem with bomber load-outs. However, to cut a long detective story short: The SH3 program has an intermittent problem with radar warning ranges set to 30,000 metres or more. For NYGM, this affects the Metox and Tunis radar-warning receivers (RWRs), which both have a range of 30,000m. Actually, in real life, both had a range of around 50,000 metres, but the range was cut down in NYGM to 30,000 metres, which is the limit of the real-life horizon. The extraordinary thing about this program bug is that (with NYGM), it only affects the standard Liberator. The Sunderland and Halifax, both long-range bombers with the same radar equipment in November 1944, are unaffected. (Typical of a possible memory leak in the code, devs please note.) If the ranges for Metox and Tunis are set to 24,000m, the air crash problem disappears completely. After extensive testing, it is clear that this range is easily sufficient for the U-boat to dive to avoid an aircraft, even at time-compression = 1024. I intend to release a patch, with other small fixes, for NYGM shortly. Anyone who has used in the past the ancient (2005) Stiebler 'Radar_Detection' fix (a stand-alone mod, also used in the old RUb mod from which other super-mods have been derived subsequently) should use a hex editor (or, better, SH3 Mini-Tweaker or SH3 File Analyzer, or S3DEditor) to change both MaximumRange and PreciseRange for the Metox and Tunis RWRs to 24000 from the existing 30000 (thus four changes in all). The affected file is silenthunteriii\data\library\sensors.dat. I have also checked GWX3 RWR ranges. These are set for Metox to 22000 and for Tunis to 25000, so probably users of GWX do not need to make any changes. Stiebler. __________________ NYGM Tonnage War Mod - More than a mod: it's an experience!

07-15-09, 12:44 PM	#2
coronas Ace of the Deep Join Date: May 2005 Location: Asturias, España Posts: 1,168 Downloads: 39 Uploads: 0	Thanks for the tip, Stiebler. __________________

07-15-09, 03:43 PM	#3
NGT Lieutenant Join Date: Sep 2006 Posts: 259 Downloads: 290 Uploads: 3	Thanks Sir, I meet hard difficulties to write in English, and my presence in the forum is limited due to this problem. But, I seize this opportunity to express my warmest thanks for your global contribution for the improvement of playability and diversity of this game. Your reports were always clear, well draw up, analytical as much are necessary, interesting and useful TO EVERYBODY. You are not limited to improve NYGM, but you share the fruit of your research with the whole community, and you offer your knowledge for improving the technical level of each one from us, and, by the same way, for improving the entertainment we need after a day in real life’s problems. All that, without any behavioral conflict, without any offense. Your attitude is an example to follow, here and in real life. Your stance go beyond NYGM (a very serious compilation indeed), beyond SH3. Certainly, a real gentleman.

07-15-09, 11:10 PM	#4
onelifecrisis Maverick Modder Join Date: Aug 2007 Location: England Posts: 3,895 Downloads: 65 Uploads: 3	Yeah +1 to what NGT said (minus all the flowery stuff ). __________________ Freedom of speech - priceless. For everything else there's Mastercard.

07-16-09, 01:10 AM	#5
Kpt. Lehmann GWX Project Director Join Date: Jul 2005 Location: The Republic of Texas Posts: 6,993 Downloads: 124 Uploads: 0	I concur with the findings posted on this thread... and can now confirm (although to a slightly lesser degree) Stiebler's assessments, though it has been a rather long time since I've observed the situation referred to regarding inexplicably downed aircraft. (As I recall : ) When the RWR ranges were set farther away to match with historical values, I witnessed a handful of oddities that in retrospect may have been related to overlong RWR ranges set in relation to SH3's limitations. However, they were infrequent enough... that I simply didn't recognize them as a consistent problem... and/or thought maybe it was just my own machine/installation. It simply wasn't 'clear' to me at any rate. Rubini attempted to persuade me to lower them for other reasons which I understand better now... and which now compliment this situation as a possible fix. Currently I use his "Unofficial RWR Range" fix which reduces RWR ranges in GWX further than those accurately listed by Stiebler above... and is accurately described by Rubini as (paraphrasing) 'Making it no longer possible to have a cup of coffee before diving' when under attack by air. If I can take a bit of liberty here without naming sources... (please feel free to refute... its late and I should be in bed at the time I am writing this.) ... It is my feeling/understanding based on anecdotal accounts of U-boats under air attack... that RWR units were far from 100% effective... and that ranges given are likely a total maximum that could be expected in perfect conditions. In my experience with Rubini's RWR mod, it seems to me that recent events during my patrols... now more closely match those anecdotal accounts. Secondarily, I have found that the associated modifications (optional inclusions IIRC) to AI rocket-firing solutions work as advertaised... ...AND something that I'm not sure even Rubini is aware of... Using the same mod, units using aircraft bomb and DC's have fewer instances of "failures to drop ordnance" on first passes unlike before. Aircraft will still make multiple passes... but not quite as many... given that ordnance is dropped sooner and generally close enough to be a threat to the boat. (Either causing actual damage or at least rocking the boat via near miss.) It seems as if the ballistic solutions plotted by the game are now a bit more 'loosened-up.' This condition should also result in fewer aircraft being shot down by U-boats in-game by those who DO choose to engage... and thereby further encouraging players to dive as opposed to engage... given that the consequences are a bit more costly. (Fewer repeating passes by aircraft and greater chance of damage by ordnance dropped during first pass.) It wasn't my intention to derail the thread by going OT with rockets and bombs... but I think that little package of mods built by Rubini is bearing out to be something we should maybe consider/request to be "official" augmentations to the GWX and NYGM mods respectively. __________________ www.thegreywolves.com All you need is good men. - Heinrich Lehmann-Willenbrock Last edited by Kpt. Lehmann; 07-16-09 at 01:30 AM.

07-16-09, 07:13 AM	#6
irish1958 Ocean Warrior Join Date: Apr 2005 Location: Carmel, Indiana Posts: 3,250 Downloads: 320 Uploads: 11	Kpt. L, I fully agree; Rubini's mod is terrific and the pyrotechnics are spectacular. __________________ Irish1958

07-17-09, 11:48 AM	#7
Stiebler Fuel Supplier Join Date: Oct 2005 Location: London, UK Posts: 1,237 Downloads: 29 Uploads: 4	The patch that I promised above for NYGM, together with other minor fixes, is now available as NYGM 3.3 from my signature below. Or see: http://www.subsim.com/radioroom/show...&postcount=247 Interesting observations from Kpt Lehmann. One can mimick unreliability in the RWR by providing a longer 'sweep period' in sensors.dat for each item, Metox, Borkum, Naxos and Tunis. Thanks for the kind words, NGT and OLC. Stiebler. __________________ NYGM Tonnage War Mod - More than a mod: it's an experience!