SRL Archive - SRL Suggestions

InForesterRandom was broken, obviously causing SolveForester to fail.

Current InForesterRandom:

function InForesterRandom: Boolean;
var
  Count, fX, fY, X, Y: Integer;
  Pts: TPointArray;
begin
  if (FindColor(X, Y, 60909, MMCX - 20, MMCY - 20, MMCX + 20, MMCY + 20)) then  //Exists NPC dots
    if (not(InBlack)) then                                                      //No black in Forester
    begin
      FindColorsTolerance(Pts, 60909, MMX1, MMY1, MMX2, MMY2, 0);
      Count := Length(Pts);
      if (Count < 10) then Exit;                                                //More than 10 NPC dots (pheasants)
      if (CountColorTolerance(clWhite, MMX1, MMY1, MMX2, MMY2, 50) < 20) then   //No white walls
      begin
        ForesterCenter(fX, fY);
        {$ifdef Simba}
        FilterPointsDist(pts,0,50,fx,fy);
        {$else}
        FilterPointsPie(Pts, 0.0, 360.0, 0.0, 50.0, fX, fY);
        {$endif}
        Result := (Length(Pts) = Count);                                        //All NPC's concentrated in centre
      end;
    end;
end;                                                                         

Colors were outdated, and the NPC dots simply are not focused around the center anymore.

Fixed:

function InForesterRandom: Boolean;
var
  Count, X, Y: Integer;
  Pts: TPointArray;
begin
  if (FindColor(X, Y, 62965, MMCX - 20, MMCY - 20, MMCX + 20, MMCY + 20)) then  //Exists NPC dots
    if (not(InBlack)) then                                                      //No black in Forester
    begin
      FindColorsTolerance(Pts, 62965, MMX1, MMY1, MMX2, MMY2, 0);
      Count := Length(Pts);
      if (Count < 10) then Exit;                                                //More than 10 NPC dots (pheasants)
      Result := (Length(Pts) = Count);                                        
    end;
end;                                              

Succesful debug:

Compiled succesfully in 624 ms.
SRL Compiled in 0 msec
Forester: We should talk to the freaky forester.
Forester: Kill The 3 Tailed Pheasant.
Forester: Killed Pheasant.
Forester: Leaving Forester.
Forester: Forester random solved.
Successfully executed.

Applied. I really wonder why there was that ifdef simba filterpointsdist else filterpointspie, though.

Did you check it for false positives too? Or is that very unlikely?

Also, been out for a bit; is 62965 the new background colour for NPCs?

From: Wizzup?
Applied. I really wonder why there was that ifdef simba filterpointsdist else filterpointspie, though.

Did you check it for false positives too? Or is that very unlikely?

Also, been out for a bit; is 62965 the new background colour for NPCs?

I ran it once when it wasn't in the random and it returned negative, but that's it.

62965 is just the most common color for NPC dots I could find.

Thanks for the fast commit :)

if (Count < 10) then Exit;                                                //More than 10 NPC dots (pheasants)
Result := (Length(Pts) = Count); 

can be made into

Result := (Count > 10);                                                //More than 10 NPC dots (pheasants)

From: Zyt3x

if (Count < 10) then Exit;                                                //More than 10 NPC dots (pheasants)
Result := (Length(Pts) = Count); 

can be made into

Result := (Count > 10);                                                //More than 10 NPC dots (pheasants)

Result := (Count >= 10);

;)

What are you talking about? The NPCs are still centre oriented…

Also, the npc dot colour didn't change. It's still 60909.

Your function would return true if there was an npc around the centre of the minimap and also if there was more than 10 npcs on the minimap.

Your function even returned true when I was in the rat pit, below varrock.

A reason why it could have been failing previously was the white wall counting mechanism. At the end of August, I changed the count of 20 to 40, to account players and whatnot, and it worked, but since I didn't have an SVN account at that time, I couldn't commit it so I posted it but it was never committed and was forgotten of.

I made a new function:

function InForesterRandom: Boolean;
var
  Count, fX, fY: Integer;
  Pts: TPointArray;
begin
  if (CountDots('npc') > 10) and (not InBlack) and
    (CountColorTolerance(clWhite, MMX1, MMY1, MMX2, MMY2, 50) < 40) then //White walls
  begin
    Pts := GetMiniMapDots('npc');
    Count := Length(Pts);
    ForesterCenter(fX, fY);
    {$ifdef Simba}
      FilterPointsDist(pts,0,50,fx,fy);
    {$else}
      FilterPointsPie(Pts, 0.0, 360.0, 0.0, 50.0, fX, fY);
    {$endif}
    Result := (Length(Pts) = count);
  end;
end;

Which works beautifully. Any last words before I commit this?

FYI, I was getting false positives in Varrock West Bank with the new Forester detection.

Here is a more efficient

function InForesterRandom: Boolean;
var
  Count, fX, fY: Integer;
  Pts: TPointArray;
begin
  if (CountColorTolerance(15726327, MMX1, MMY1, MMX2, MMY2, 30) < 40) then //White walls
    if (not (InBlack)) then
    begin
      Pts := GetMiniMapDots('npc');
      Count := Length(Pts);
      if (Count < 10) then
        Exit;
      if (not (ForesterCenter(fX, fY))) then
        Exit;
      {$ifdef Simba}
        FilterPointsDist(pts,0,50,fx,fy);
      {$else}
        FilterPointsPie(Pts, 0.0, 360.0, 0.0, 50.0, fX, fY);
      {$endif}
      Result := (Length(Pts) = count);
    end;
end;

Short circuit evaluation.
More accurate wall color average lower tolerance.
Only calls the NPC dot collection one time.

We probably need another layer to the function to avoid false detections like what Tara's saying.

Also you can't do Result := Count > 9, it's not the same ;) the NPC dots before and after the filter need to be the same.

function InForesterRandom: Boolean;
var
  Count, fX, fY, CTS: Integer;
  Hue, Sat: Extended;
  Pts: TPointArray;
begin
  GetColorspeed2Modifiers(Hue, Sat);
  CTS := GetColorToleranceSpeed;
  if (CountColorTolerance(15726327, MMX1, MMY1, MMX2, MMY2, 30) < 40) and //White walls
    (not (InBlack)) then
    begin
      Pts := GetMiniMapDots('npc');
      Count := Length(Pts);
      if (Count < 10) then
        Exit;
      if (not (ForesterCenter(fX, fY))) then
        Exit;
      {$ifdef Simba}
        FilterPointsDist(pts,0,50,fx,fy);
      {$else}
        FilterPointsPie(Pts, 0.0, 360.0, 0.0, 50.0, fX, fY);
      {$endif}
      if (Length(Pts) = count) then
      begin
        SetColorToleranceSpeed(2);
        SetColorspeed2Modifiers(0.67, 1.68);
        if (PercentColorMM(1397036, 8) > 10) then
          Result := True;
      end;
  end;
  SetColorToleranceSpeed(CTS);
  SetColorspeed2Modifiers(Hue, Sat);
end;

That code also checks to see how much of the minimap is covered in trees. Returns true if it's >10%.

Committing this now.

Why not use short circuit?

if (CountColorTolerance(15726327, MMX1, MMY1, MMX2, MMY2, 30) < 40) and //White walls
    (not (InBlack)) then

->

if (CountColorTolerance(15726327, MMX1, MMY1, MMX2, MMY2, 30) < 40) then //White walls
    if (not (InBlack)) then

From: IceFire908
Why not use short circuit?

if (CountColorTolerance(15726327, MMX1, MMY1, MMX2, MMY2, 30) < 40) and //White walls
    (not (InBlack)) then

->

if (CountColorTolerance(15726327, MMX1, MMY1, MMX2, MMY2, 30) < 40) then //White walls
    if (not (InBlack)) then

What's the difference? I just like using less if statements if they aren't needed.

From: TRiLeZ
What's the difference? I just like using less if statements if they aren't needed.

It's important to stagger functions for efficiency, is you use and statements it will needlessly call every function in the statement if one of them fails already.

From: IceFire908
It's important to stagger functions for efficiency, is you use and statements it will needlessly call every function in the statement if one of them fails already.

program new;

function a: boolean;
begin
  Writeln('a');
  Result := False;
end;

function b: boolean;
begin
  Writeln('b');
  Result := False;
end;

begin
  if (a and b) then
    Writeln('k');
end.

If one of the conditions is false, it will exit the statement.

Very interesting

program new;

function a: boolean;
begin
  Wait(1000);
  Writeln('a');
  Result := False;
end;

function b: boolean;
begin
  Wait(1000);
  Writeln('b');
  Result := False;
end;

function c: Boolean;
begin
  Wait(1000);
  Writeln('c');
  Result := False;
end;

var
  T: Integer;

begin
  T := GetSystemTime;
  if (a and b and c) then
    Writeln('k');
  WriteLn(GetSystemTime - T);
end.

Simba :

Compiled succesfully in 15 ms.
a
1107
Successfully executed.

SCAR 3.25

Successfully compiled (78 ms)
a
998
Successfully executed (998 ms)

but SCAR 3.23 and older

Successfully compiled (2568 ms)
a
b
c
2995
Successfully executed

Nice find.

Not going to lie, Squancy's main just got this random last night as I was going to bed and it failed so he asked me to fix it quick. So I fixed it, quick.

From: TRiLeZ
What are you talking about? The NPCs are still centre oriented…

I had it fail multiple times due to not finding NPC's before I removed the FilterPoints**.

From: TRiLeZ

Also, the npc dot colour didn't change. It's still 60909.

From: TRiLeZ

I made a new function:

function InForesterRandom: Boolean;
var
  Count, fX, fY: Integer;
  Pts: TPointArray;
begin
  if (CountDots('npc') > 10) and (not InBlack) and
    (CountColorTolerance(clWhite, MMX1, MMY1, MMX2, MMY2, 50) < 40) then //White walls
  begin
    Pts := GetMiniMapDots('npc');
    Count := Length(Pts);
    ForesterCenter(fX, fY);
    {$ifdef Simba}
      FilterPointsDist(pts,0,50,fx,fy);
    {$else}
      FilterPointsPie(Pts, 0.0, 360.0, 0.0, 50.0, fX, fY);
    {$endif}
    Result := (Length(Pts) = count);
  end;
end;

Which works beautifully. Any last words before I commit this?

That's the original one jumbled into a single if..then with 20 changed to 40?

Nice to see people have the time to hop on this, but no one else seems to have the time to actually fix a random..

From: NCDS
Nice to see people have the time to hop on this, but no one else seems to have the time to actually fix a random..

People are working on fixing the random solvers…

We like to fix randoms by making sure the "fixes" actually work..

Geez guys, he's fixed more randoms in the last month that others have in the last year. Cut him some slack.

The only reason "some" of the randoms are working is because of me and someone else. And maybe a couple of other people that keep sending him randoms.

But i guess most of ya'll forgot that he has MSI on his hands too.

And of course everyone just bashes his fixes, but they don't do anything to help fix them. :/

I believe he has fixed 5? or maybe a little more.

Anyways, keep up the work NCDS, your fix worked for me.

From: Squancy

Anyways, keep up the work NCDS, your fix worked for me.

Thanks, but no big deal.

I messed up. I should have just waited till morning to post it so I could have tested it better, and been able to think more clearly.

Anyway, I'm over it.

From: TRiLeZ
What are you talking about? The NPCs are still centre oriented…

Also, the npc dot colour didn't change. It's still 60909.

Your function would return true if there was an npc around the centre of the minimap and also if there was more than 10 npcs on the minimap.

Your function even returned true when I was in the rat pit, below varrock.

A reason why it could have been failing previously was the white wall counting mechanism. At the end of August, I changed the count of 20 to 40, to account players and whatnot, and it worked, but since I didn't have an SVN account at that time, I couldn't commit it so I posted it but it was never committed and was forgotten of.

I made a new function:

function InForesterRandom: Boolean;
var
  Count, fX, fY: Integer;
  Pts: TPointArray;
begin
  if (CountDots('npc') > 10) and (not InBlack) and
    (CountColorTolerance(clWhite, MMX1, MMY1, MMX2, MMY2, 50) < 40) then //White walls
  begin
    Pts := GetMiniMapDots('npc');
    Count := Length(Pts);
    ForesterCenter(fX, fY);
    {$ifdef Simba}
      FilterPointsDist(pts,0,50,fx,fy);
    {$else}
      FilterPointsPie(Pts, 0.0, 360.0, 0.0, 50.0, fX, fY);
    {$endif}
    Result := (Length(Pts) = count);
  end;
end;

Which works beautifully. Any last words before I commit this?

Yeah. Once again, why use FilterPointsDist in Simba, and FilterPointsPie in SCAR?

From: Wizzup?
Yeah. Once again, why use FilterPointsDist in Simba, and FilterPointsPie in SCAR?

I wasn't the one who came up with the idea of doing that, so I just left it like that.

From: Wizzup?
Yeah. Once again, why use FilterPointsDist in Simba, and FilterPointsPie in SCAR?

Ask Raymond, lol?