I've altered the Move values at the "WALKTO" routine, making sure they will be the same as the player position when the walking starts

At least I fixed the crash caused when a route failed

The values that my MYPOS commands gives, do seem to be in line with what Dijkstra throws me, so everything looks in order, but that does not come out in the way Irravonia walks... So something must be wrong with the way Kthura calculates back the values Dijkstra gives it into the actor... Which is odd at best....

One "null" reference in case of impossible walks

I've added a debug line which should give more confirmation about matters

I've made a quick calculation of the ending value, and at least that value appears to be in order.... The other walk values do also appear to make sense to me

Let's find out more about this matter

yeah, that did fix stuff, but not really in the way expected, so I still got a lot of work to do

Oh yeah, forgot to tell you, Kitty enjoyed her meal by now :P

Of course, i gotta test this.... :-/

I think I fixed this issue.... Now all updates will happen AFTER the actors have moved, and my means of optimization ONLY if they have moved.

I think I solved the mystery... I think the automapper is to blame for this, which immediately activates on ANY sort of modification to any Kthura object during drawing, which is PRECISELY what the moving engine does.... Now having these remappings would only lead to slowdowns anyway, so I can best make sure they are turned off during the drawing iteration

The path was correctly generated, so the error does not lie there.... However, now I get an error of the most dubious kind, clearly thrown by .NET itself, and which should not be possible no matter how you put it.... I wonder how or where this error was caused....

Oh, and I missed a C# parse error too... that also explains....

Let's do that again, then

Forgot to save my source code prior to calling the compiler

Neither success nor failure are given... This would denote that the Dijkstra linkup code was not called.... That would be the only possibility left... however the earlier out-of-bounds errors given earlier denote that the request was done, and those errors are impossible without that call... I'm at a loss now....

Let's test

Let's narrow down the possibilities, after all when you eliminate all the impossible, whatever remains, however improbable, must be the truth, after all... This debug line can rule out 50% of the now open possible scenarios to the impossible, so only 50% that can be true will then remain.... That is a huge progression, doncha think?

Apparently, that was not the only thing that wasn't right

Of course, a little test will have to confirm I'm right or not....

I think I solved the issue why little Irravonia didn't walk.... Not only the "to" coords had to be adepted to the blockmap grid... The "from" coordinates had to be done so, as well

At least due to the earlier "out-of-bounds" errors thrown by Dijkstra, I can at least be rule out the possibility that Dijkstra wasn't called at all, and you know the saying: "When you eliminate all the impossible, whatever remains, however improbable, must be the truth!"

At least the "out of bounds" protection is operational... Now I gotta find out why little Irravonia doesn't walk, when instructed to....

In the meantime, I'll let my builder scripts do their work, so I can immediately start testing while Kitty is eating... ;)

Kitty (my cat) is begging me to fill her food bowl, so I guess I should do that.... It's about time too, so she has the right to do so....

The Dijkstra out-of-bounds protector needed a "return" command, since it was missing the code causing the error was still executed...

All fixes ignored???

A little NIL translation issue

Indeed, Dijkstra and Kthura were not in one mind when it came to how to use booleans for passible or impassible areas.... I made an extra routine in Kthura to make the reverse, but I also kept the original.... This to prevent compatibility issues with other stuff....

In the meantime my backup has finished so I can get back to "business as usual" mode

What I thought is true, Dijkstra did have an out-of-bounds protection during the route calculation, but not during the check of the starting and end coordinates.

If you request starting and ending coordinates outside the bouds of the blockmap, the path calculation will now always result into a failed empty route

A few things are clear, though

1. The coordinates are correctly adapted to the blockmap grid, so at least that is not the issue
2. It is also clear that the protection for walking out of bounds is not really operative... The test program did make the route touch the edges (on purpose) to see if that would properly block things.... So I must fix Dijkstra itself to always reject out-of-bounds walking requests
3. I do not know if Irravonia not walking is an actual failure in the route calculation, or the blockmap, but I think I got an idea on this though, as I think Dijkstra and blockmap have reversed meanings on True and False, but I need to verify that one

An extra debug line will have to provide me some information on the reason WHY!

This is a very common bug I had with the pathfinder in the original Dyrt as well, also when I was on Star Story, and my 63 fires of lung.... Only TFT REVAMPED didn't suffer here, but that was only because TFT REVAMPED and Star Story use the same engine, so the issue was fixed before that project began XD

Wherever I click NOTHING happens at all...

Take II

Illegal group call

Let's see what happens.. Take I

And now the command to make Irravonia actually walk has been given, meaning she should walk now to the location the player directs her to, *if* in fact she can reach it... Please note, due to the way the blockmap has been set up, Little Irravonia will keep herself to a 32x32 grid

So I can start my test anyway, as I don't wanna waste time waiting for my backupper to finish its job (and I really need to fix some issues with that tool asap).

Well luckily Powershell grants me a few quick commands to "void" that....

Unfortunately my backup tool got a bit stuck in the same window from where I run my building scripts... I hate it when that happens.... :-/

And let's see what that'll do!

For now I've limited myself to making sure the system understand where you want the hero to walk to

I've Made a "dummy" field... It does nothing yet, but at least this makes it easier on me to make 100% sure I can easily do alternate movements, like clickable objects, and that is very important for NPCs, savespots, signs and other things you can find in your everyday RPG dungeon or city.

I'll make a few worksouts in preparation to use Dijkstra though.... I got some time before I really must quit my session today

I needed a little break! Since the walking around routine is also more complex than just setting up a Dijkstra calculation whenever the player clicks something (as there are also clickable objects in the field), this may be something that won't be done until after the weekend, but when I say such things, you never know...

You should all know by know how quickly I can change my mind when it comes to these things....

I did a test and no parse errors came up, which is so far a good thing

All i can do now is to make sure all parse errors are taken care of... As the game doesn't yet have any controles allowing you to move, the Dijkstra routines are not called or anything yet.

#27 (And it doesn't look important, as this causes under-the-hood-issues, the player may never see, but memory leaks (even when they are resolved as soon as the game ends) are always a bad thing)

And the NIL link code has been created

Or rather a 3rd as a WalkToSpot was also there

The C# WalkToCoords has been written.... That is half of the contract

Of course, now I need to write the proper link code for Bubble, which is in this case for NALA

Moving in general and also in accordance with Dijkstra

Kthura is now able to calculate the route to walk over the Kthura Map, thanks to the underlying Dijkstra unit....
I must note though, that the route will be calculated, but the result will be ignored as I do not have th proper walking update stuff available yet.... (can't do all in once you know).

(This will require me to include Dijkstra also in the Kthura Map Editor source code, but that is a later concern).

I've now set the path finding variables to work with my own Dijkstra module

Dijkstra Has now been linked to NALA, so I can use it in Kthura

I'v deliberately created a track that was impossible to get from start from finish to see if this would indeed get a 0 path and a "false" for success rate... This happens to be the case

Time is however not on my side to actually attach this code to Kthura and to Bubble, but that doesn't matter right now... It works, and that's important ;)

And how come I never realized how simple this algorithm was..... :-/

I've been working things out on the Nodes parents, and NOW it works.... And both the demo map and the breakdown confirm that.... WOOOHOOOO!

.git folder form JCR6 in the test project

But seeing this mapped out, with a complete breakdown list, this no longer makes sense.... Some nodes have been correctly generated, yes, but been linked to the wrong parent....

I could not yet see the actual route, but.... What I did see was that the route was calculated to be 24 units long, and guess what, when I walk the route manually as a human, I come to the same calculation.... This could already be a good indication, the Dijkstra routine works....

Testing stuff

But as I'm now just in testing stuff, not gonna do that, and let's just add JCR6 so I can get this over with!

Perhaps I need to revise qstr, as I wonder why a quickstring routine needs JCR6 support ;_/

Error fixing

Quick Test Program set up

Incoming phone call delayed me

Before I can test this I need to write some test code though.....

Frankly speaking I SHOULD have set up a Dijkstra based pathfinder now, however if it's robuust or working at all.... I can't tell, so a test is in order.

Created empty Dijkstra.cs file

And with that I think I removed everything that can cause me some trouble

Quick form calls

Backups also not needed

I won't use GTK in these tests

Unneeded documents

So I removed at least the link the git-crap

Now I did link in the entire TrickyUnits folder, but I surely won't be needing that XD

For now I will focus on the pathfinder alone as a separate project.... If succesful I got my own pathfinder which I can later on also export to other languages, so there's a lot to win for me here :P

Wait a minute.... I think I can create a kind of Dijkstra based pathfinder myself now.... The step by step description I see now is the first one set up in a way that maks sense.... Perfect....

I've made a request in AnnSerba/A-star#1 ... This can prevent that I may have to remove the code Ann Serba provided has to be removed in the future, wasting a lot of my time in the process...

I've studied the pathfinder routine for C# I found... They do not look too difficult to understand and use, however they came without any license terms, and that could be a bit of a bummer. Now NALA has been released under a GPL license, but Kthura as a library has not been so, I really need to work this out in a way licenses cannot be an issue.

The debug line for faders.... I don't think I need that anymore, as it appears to be working

APpears to work

Let's try that again, shall we?

SASKIA typo

Of course, I gotta try if this all works

I've made it possible for the autoscroller to force being executed. This will allow me to make sure SASKIA scripts can force the camera to the correct position if needed

I've created a QuickNIL tool which can help me to make prototyping possible ;)

Okay, I guess it was either the visible parameter or the alpha parameter, anyway, I altered them both, and that fixed the issue, case closed :P

Let's see....

It will still be a great mystery to sort out why Irravonia doesn't show.... Perhaps the core, perhaps the MonoGame driver.... It's always a bit of an issue

And take XXIX

Run build script

And the Kthura core will also copy the spawn points dominance value into the actor... That is very extremely important, and could also be the main reason why little Irravonia didn't appear

Made sure the SASKIA script will remap

Almost there, but the autoscroll had to activate immediately and not after Irravonia speaking + Irravonia isn't seen.... The latter could just be a remap issue

Take XXVIII

Fine tuning

Take XXVII

Ah.... That explains.... I think

The error indicates something that is NOT right....

Take XXVI

Did I fix this?

It appears the leader isn't recognized for some odd reason

Take XXV

extra debug must confirm this and give more info

No frames loaded

Take XXVI

More info

I hate the NIL bug that ignores the keyword "nil" when it's the final word of the line

And Take XXV

Take XXIII and XXIV

I keep forgetting that "each" requires the tables library, which is not loaded when internal scripts are being processed :(

Another vs sync and debug run, then ,eh?

Index out of array range, eh? (C# error).

Take XXII

At least I fixed another case error, so I guess that other fix is also a real fix

I think I fixed that.... (or not)

Take XXI

Maybe this fixes it?

Right, the bug is that the picture file to be attached to the player sprite is never properly defined.... Good to know....

Take XX

Will this answer a few questions?

This all looks a bit obscure, so I hope I can get to the bottom of this...
At least I know that the texture loader is to blame

Take XIX

I've added a way to traceback C# as I need to look at that....

Let's prepare for running in VisualStudio

I'm gonna need VisualStudio to nail this one... But since this is affecting the core of Kthura, I guess I'll need VisualStudio to fix all this anyway

I've added an extra line which must tell me

I need to sort out if older Kthura maps do fall through the garbage collector or not... Some information the debugger gave me, make me think not....

Take XVIII

It does appear to confirm this takes place in the kthura drawing algorithms, and that will make it a rough ride to pinpoint the source of the bug....

Of course you realize I fixed this now!

Stacking overflow due to bad C# overloading.... Yes, I'm a first class moron!

Not quite sure what is happening, but everything is freezing, even the music freezes... This is really really bad!

Take XVII

I've expanded some data... hopefully this gives some answers (I'm afraid not, but hey, whatever I can narrow down is good, right?)

Unfortunately the traceback information is too vague

This will be one of the harder errors to track down.... This information is too minimal to do something with, however the traceback can at least show me which call from NIL is responsible, and hopefully this can narrow down my options.

Take XVI

I fixed that

SAKSKIA requires strings to be quoted! Just like normal NIL and Lua, dumbo!

Take XV

Well I fixed that

Because I forgot to set the "spot" parameter in the SASKIA Script

What I need to figure out is why the "spot" parameter is 'nil'....

And I made sure the mouse pointer wouldn't be darker in the field. Not that important, but hey I saw that this was not right, so why not set it right before I forget it! ;)

What I can at least tell is that the part without player actors appears to work again

Take XVI

Did I fix it now?

Oh, crap!

Take XV

I think I fixed it (hard to tell)

The game demands a player, which the prison can (initally) not give

Take XIV

Case errors

Take XIII

Or at least, I think so....

But to the Field flower....

Well the error I get now does at least no longer refer to the Kthura link code....

Well we get to take MCMLXXV or not?

Take XII

A leftover from old code

The trouble doesn't end there... of course

Well at least that one is fixed

Take XI will tell!

Did I fix it?

Or not?

right "nil" appears to be the parent, which I actually already expected, but at least that's confirmed now.... This is gonna be some hell to debug, I'm afraid

Take X

Expanded error message so I can (hopefully) get a better view on what exactly goes wrong

Link script chunk properly named

A break... I need to clear my brain a little

"None or incorrect parent given!" Now this will be a best to debug

Take IX

Indexing error

GRRRRR!

Take VIII

A few random occuranaces

Take VII

So this time it DOES count as a fix

The "Unknown term" error that pops up thanks to this message confirmed my earlier statement... however this was a var I forgot to rename

VI

THAT is NOT a fix... something that ain't broken, can't be fixed.... At least I hope to prevent the false crash errors this way

I've wrote some voidance code which hopefully gets me around the (incorrect) dupe identifier error

It appears all link code errors have been accounted for

Take V

Bad function type

(And I've only just begun, I'm afraid).....