Progress

[GoodGuy]

ORIGINAL: Arjuna

Good catch. The world needs nitpickers. [:)]

TT3652 - Estab - Correct grammer in description of Panther Tank

Good thing you don't have to raise a new TT for correcting bad grammar in a TT entry [;)]. That would suck, eh? j/k [:D]

[Arjuna]

[:)]

[Deathtreader]

Howdy,

Got time for a mini-update??
Thanks!

Rob.[:)]

[Arjuna]

Well I arrived back from my Bulge tour two weeks ago and promtly did my ankle in walking the dog up on the ridge. I have been working day and night writing new code to automate the construction of bridges - a key task if the German AI is to be aggressive enough. I'm just about to test it right this minute. I'll be back. [:)]

[06 Maestro]

God speed.

[GShock]

Arjuna is there a list of features for Botb? I mean...what's in when compared to httr and cota. 

[Llyranor]

There's a bunch of new awesome stuff being shown in this thread http://www.matrixgames.com/forums/tm.asp?m=1723551

[Arjuna]

Hi all,
 
Today's the 64th anniversary of the start of the Battle of the Bulge. So it's fitting I post a progress report on BFTB to mark the occasion.
 
I am currently stepping through and tidying up the new automated bridge building code. This code has impacted heavily on the scheduling code, particularly the slippage and cribbage, and the Move doctrine. To accomodate for all the cases it's been necessary to overhaul the code that plans Moves, such that we now split the assinged forceGroup into a set of subGroups based on whether they can reach the objective with or without a bridge being built. Needless to say this has complicated the AI code somewhat. But it now looks pretty spiff as it automatically determines where to build the bridges, schedules separate tasks to move each of the subGroups, assigns eng units to construct bridges and other units to defend nearby. It even merges independent groups when it can get the chance to recombine the force.
 
One of the byproducts is that I have managed to streamline the scheduling code. Move tasks that have not reached their objectives are extended rather than expiring and forcing an unnessary replan. This makes for a smarter AI as well as reducing processing loads.
 
Paul got our autotesters to work under Vista, so we ran our first run yesterday. It immediately logged over forty asserts in just over an hour. So there is still a bit to tidy up before I can put a new build for the beta testers. Hopefully we can get a new build out by the end of this week. Then we can get some screen dumps posted here highlighting the new features.

[06 Maestro]

ORIGINAL: Arjuna

One of the byproducts is that I have managed to streamline the scheduling code. Move tasks that have not reached their objectives are extended rather than expiring and forcing an unnessary replan. This makes for a smarter AI as well as reducing processing loads.

That is a very nice touch. Too bad it was not ready for this winter break, but I'm sure the wait will be worth it.

Many of us appreciate you maintaining your standards-its not a common trait these days.

[BigDuke66]

Nice to hear is's going forward.
Would be enough if we get our hands on it to the 65th anniversary.

[Sybma]

Indeed a pity it isn't available this winter holidays, but isn't it possible to make the (parts of) manuals available for download? Would make some great reading next to the fireplace!![:D]
 
 

[Arjuna]

The manual won't be ready for a while yet. We still have to incorporate some UI changes that will affect the screen dumps included in the manual. So I'm afraid you will have to find some other holiday reading material. We are working flat out right now. As sson as we can we'll release some more info.

[ETF]

How has the new beta build worked out?
Do you think maybe a Summer release is likely at this stage? Just curious

[Arjuna]

I've been taking a much needed XMas/NY break. I have still to rework the scheduling code to accomodate the new automated bridging code. So we are yet to put out a new build - hopefully next week. As to when BFTB will be released, well I am loathe to make a prediction till after I have finished this reworking of the scheduling code. So I'll be in a better space to estimate once that is done.

[hank]

No problem here.  A few weeks ago I played several HttR scenario's and this weekend I got another one in.  I had forgot how much fun HttR is.  I'm having a blast playing through Market Garden again. 
 
Over the Holidays I also cranked up HPS's Bulge for a few short scenarios and I'm going to play a couple of the user scenarios of the Bulge with the HttR engine.  I downloaded them a long time ago but never played them.
 
All this in anticipation of BFtB.
 
(a bump in disguise)  [8D]

[Arjuna]

Oh happy day! I have finally got the automated bridge construction to work. What a marathon effort.

Below is a screen dump taken from Race for Bastogne #2 scenario in which by around 1900 on Day 1 the German's have successfully rebuilt two of the principal river crossings. Note thgat German motorised forces have already crossed over and are on the way to their objectives on the far bank. Note also that the German foot infantry have almost secured the skyline drive objectives on the far bank after a day of hard fighting. American forces in Hosingen are still holding out despite now being surrounded. How much longer they will last I don't know. But what I can say is that now that the bridges have been rebuilt their chances just got a whole lot worse.

[Arjuna]

Previously the AI would only rebuild bridges where the scenario designer had placed a victory objective and set the rebuild bridge flag - ie only those designated by the scenario designer. Now the AI will automatically rebuild a bridge where a force is ordered to an objective across an unbridged river provided it has a bridging capability. If it does, then the AI will determine the crossing to rebuild, schedule the approapriate construction tasks and allocate the forces accordingly.
 
The AI is smart enough to split the force and allow foot units that can cross the river or who are already across the river to head off towards the objective by themselves, leaving the motorised to defend on the near bank of the crossing while engineers construct the new bridge. Once rebuilt the force replans and the motorised elements now head off across the rebuilt bridge.

[Arjuna]

This new feature has been essential to simulate the fighting in the rugged Ardennes which is criss crossed by so many rivers. Trying to have the scenario designer second guess the need for bridge rebuilding was proving too difficult in the free flowing battles of manouvre that develop. It's time now to put out a new build for the beta testers to put this new feature through its paces. [:)]

[goodwoodrw]

G'day, well done Dave, I don't know much about programing, but the above explanation isn't straight forward so to speak, which means the coding of this feature sounds like an exceptional feat. Well done sir you deserve all the fruits you get from your labour.
Ron

[Arjuna]

Actually coding this feature was a big challenge. So much of the scheduling code had to change to get it to work. The previously fairly simple Move doctrine has become a lot more complex. It now splits up the forces for a Move order into three categories - those that can reach the objective right now, those that can't but could if a bridge was built and those that cannot regardless. It then processes each unit within each group, determining if they should move directly to the objective, stay put or move to an interim crossing point. It handles independent moves where units are too far apart and can merge these later. A Move plan can now have a dozen or more tasks, including interimMoves, interimConstructs, interimDefends, feederMoves, finalMove, not to mention the usual complement of firebase and depot tasks.
 
With this amount of tasks it became important to review the scheduling code so that the tasks could be slipped and cribbed as tasks finished or were delayed. This would have been so much easier if we had already implemented triggers - ie conditions that govern when a task can start or end. As it was I had to hard code in a lot of case specific functionality here. Hopefully in the future we can replace this with more elegant trigger code.