Dbl bfr #58

tehKaiN · 2018-05-06T16:57:02Z

fixes #57
fixes #34
fixes #59

Imposes a breaking change: pointer to simplebuffer's bitmap is changed from pManager->pBuffer to pManager->pFront and pManager->pBack.

approxit · 2018-05-07T11:00:02Z

src/ace/managers/viewport/simplebuffer.c

+			uwBoundWidth, uwBoundHeight,
+			pVPort->ubBPP, ubBitmapFlags
+		);
+		if(!pFront) {


approxit · 2018-05-07T11:01:18Z

src/ace/managers/viewport/simplebuffer.c

@@ -81,7 +186,8 @@ tSimpleBufferManager *simpleBufferCreate(
 		logWrite("Copperlist offset: %u\n", pManager->uwCopperOffset);
 	}

-	simpleBufferSetBitmap(pManager, pBuffer);
+	simpleBufferSetFront(pManager, pFront);
+	simpleBufferSetBack(pManager, pBack == 0 ? pFront : pBack);


Explicit == 0 is not needed.
pBack ? pBack : pFront

approxit · 2018-05-07T11:03:20Z

src/ace/managers/viewport/simplebuffer.c

-			(6+2*pManager->sCommon.pVPort->ubBPP)*sizeof(tCopCmd)
-		);
+	if(pManager->pCameraManager && isCameraCreated) {
+		cameraDestroy(pManager->pCameraManager);


Accessing pManager that was freed in line 206?

approxit · 2018-05-07T11:17:11Z

I need to say NO for breaking change with removing pBuffer pointer.

Actually games explicitly use pBuffer everywhere. pBack/pFront is a hard coded pattern bound only to simplebuffer manager. Other buffer managers use pBuffer. You don't want to refactor whole game code while switching on/from simplebuffer manager.

Let's keep the old pointer to keep sanity, and pFront/pBack as additional pointers.

tehKaiN · 2018-05-07T11:29:07Z

Okay, hold on. This braking change is minor and I've mentioned it only so that you can effortlessly update code in your prods. Keep in mind that we're not versioning ACE yet, so unless we're in semver 1.0 expect stuff to change a lot.

I expect that ACE will have little to no backwards compatibility between versions for sake of speed, because there are lots of stuff in code which were written in a very poor way. Best approaches are not yet known, so I'm not going to support wrong ones.

This change is for the better, since for single-buffered screens pFront and pBack will be the same, and if you work on pBack then you can switch to double buffered stuff without changing anything else except for one creation tag.

Also, recent OS disabling and GCC pull requests probably broke your code anyway, so idk why you're getting mad at such trivial change.

If you really don't want to change code in your games, just include ACE as submodule with HEAD set to last known commit which worked.

Regarding simplebuffer-only pattern, expect double buffering support in tile/scroll buffers too.

approxit · 2018-05-07T11:36:42Z

src/ace/managers/viewport/simplebuffer.c

@@ -187,7 +187,7 @@ tSimpleBufferManager *simpleBufferCreate(
 	}

 	simpleBufferSetFront(pManager, pFront);
-	simpleBufferSetBack(pManager, pBack == 0 ? pFront : pBack);
+	simpleBufferSetBack(pManager, pBack ? pFront : pBack);


Ommiting == 0 actually flips the logic.

fixed in 6301206

approxit

I expressed myself not correctly, sorry.

I'm not against breaking changes. What I mean is to keep engine API consistent.
Breaking out simplebuffer manager from pBuffer which is used by EVERY other buffer manager is breaking the consistence. It's API job to ease end-users with working with engine. Double buffering is an optional feature which forces the-most-simple-case (single buffered mode) to have a penalty in decision about buffer manager change.

Good API should have most common cases as defaults. That's why I want to keep pBuffer in singlebuffer manager. Optional feature is not worthy of removing that handy consistent name regardless of the manager you use.

approxit · 2018-05-07T11:37:36Z

src/ace/managers/viewport/simplebuffer.c

@@ -202,12 +202,12 @@ tSimpleBufferManager *simpleBufferCreate(
 	if(pFront) {
 		bitmapDestroy(pFront);
 	}
+	if(pManager && pManager->pCameraManager && isCameraCreated) {


It can be nested inside pManager check. Why check twice, when you can check once?

tehKaiN · 2018-05-07T15:11:54Z

We may wait with merge until rest of vport managers get double buffering support then - it will be consistent across whole engine.

From my current point of view, writing games using single buffering shouldn't be a default approach since it works in very few cases, where you're sure you will always be with your drawing before display beam. Double buffering allows you to not waste computing time on bob sorting which gets costly if you also consider z-sort. Without sorting in mind, your only concern is if you finish all blits in time of single frame, or in two frames if you target 25fps instead of 50.

scrollBuffer now disables second copBlock if not needed

previous version allowed drawing on outermost margins, which resulted in drawing stuff when it shouldn't be drawn

tehKaiN · 2019-05-07T12:33:33Z

i don't want to wait for it any longer, seems stable enough

tehKaiN requested a review from approxit May 6, 2018 16:57

approxit reviewed May 7, 2018

View reviewed changes

tehKaiN force-pushed the dbl-bfr branch from 7264d99 to 009dad2 Compare May 17, 2018 15:32

tehKaiN force-pushed the dbl-bfr branch 3 times, most recently from 5eb3d25 to 88dd170 Compare June 10, 2018 10:33

tehKaiN force-pushed the dbl-bfr branch 3 times, most recently from f2f47c5 to ec41b96 Compare June 25, 2018 11:29

tehKaiN force-pushed the dbl-bfr branch 2 times, most recently from 2e2d12a to e3d9ad3 Compare July 10, 2018 19:08

tehKaiN force-pushed the dbl-bfr branch 3 times, most recently from a6e9d25 to db49889 Compare July 15, 2018 12:34

tehKaiN force-pushed the dbl-bfr branch 4 times, most recently from cc64a3d to 1229792 Compare July 24, 2018 06:32

tehKaiN force-pushed the dbl-bfr branch 3 times, most recently from 0c5fdb8 to a7d9142 Compare September 7, 2018 18:04

tehKaiN force-pushed the master branch 2 times, most recently from d55fe9d to 57733d4 Compare September 7, 2018 18:08

tehKaiN added 6 commits April 16, 2019 19:23

Unified fn names in joy and key managers, made 4joy polling optional

fbdd3c4

Added double buffering for camera

efad313

Some stuff that asman pointed out

75fda79

fixed disabling CopBlocks, scrollBuffer tweak

007d061

scrollBuffer now disables second copBlock if not needed

perhaps it'll fix #88

5f974de

fixed tileBufferIsTileOnBuffer() region check

5ba6a43

previous version allowed drawing on outermost margins, which resulted in drawing stuff when it shouldn't be drawn

tehKaiN force-pushed the dbl-bfr branch from bf759cc to 5ba6a43 Compare April 16, 2019 17:23

tehKaiN added 13 commits April 24, 2019 19:13

minor formatting

38e0d77

parse tRGB form hex code string, tileset_conv has settable bg color

b1c37fc

reorder to prevent std::move crashes

c197d23

added helpers to main cmakelists

1fec276

this is needed for helpers

4ce920f

promoted fontGlyphWidth from static since it can be useful

fdf4891

added cmake util for font conversion, minor tool fix

1ff81a0

Merge branch 'tools-tileset_conv_bg' into dbl-bfr

c59b763

showcase builds again

48b9194

added missing include

4ae3062

added buffer scrolling test

315b5ee

fixed first line not being displayed on simple buffer

eacf039

added unimplemented notice

971784e

tehKaiN force-pushed the dbl-bfr branch from 0240269 to 971784e Compare May 4, 2019 16:37

tehKaiN added 2 commits May 4, 2019 19:12

fixed scrollBuffer not displaying y=0 line

5ec8573

fixed destroying simplebuffer in runtime by adding copBlock freeing

1e7e64e

tehKaiN force-pushed the dbl-bfr branch from abfefbd to 1e7e64e Compare May 4, 2019 18:02

tehKaiN added 3 commits May 7, 2019 13:31

obvious fix

227a3d7

increased ruler scale

9f2c2a5

added check for proper scroll buffer y-wrap

5e20ad9

tehKaiN merged commit f50d520 into master May 7, 2019

tehKaiN deleted the dbl-bfr branch July 25, 2020 07:10

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Dbl bfr #58

Dbl bfr #58

tehKaiN commented May 6, 2018 •

edited

Loading

approxit May 7, 2018

approxit May 7, 2018

approxit May 7, 2018

approxit commented May 7, 2018

tehKaiN commented May 7, 2018 •

edited

Loading

approxit May 7, 2018

tehKaiN May 7, 2018

approxit left a comment

approxit May 7, 2018

tehKaiN commented May 7, 2018

tehKaiN commented May 7, 2019

Dbl bfr #58

Dbl bfr #58

Conversation

tehKaiN commented May 6, 2018 • edited Loading

approxit May 7, 2018

Choose a reason for hiding this comment

approxit May 7, 2018

Choose a reason for hiding this comment

approxit May 7, 2018

Choose a reason for hiding this comment

approxit commented May 7, 2018

tehKaiN commented May 7, 2018 • edited Loading

approxit May 7, 2018

Choose a reason for hiding this comment

tehKaiN May 7, 2018

Choose a reason for hiding this comment

approxit left a comment

Choose a reason for hiding this comment

approxit May 7, 2018

Choose a reason for hiding this comment

tehKaiN commented May 7, 2018

tehKaiN commented May 7, 2019

tehKaiN commented May 6, 2018 •

edited

Loading

tehKaiN commented May 7, 2018 •

edited

Loading