see D4627
"Needs another fix to be tested so trust me bro ™"
ok

Apply the patch and compile the game

ok

Check that text input and multiline text are correctly clipped

I did not encounter any problems.

Apply the patch and compile the game
Run the game

ok

Change something in shaders

Changing binaries/data/mods/public/shaders/glsl/common/debug_fragment.h while the game is running

• from return vec3(alpha);
• to return vec3(1.0, 0.8, 0.0) * (1.0 - alpha);

It would crash without the patch being applied.
With the patch, a crash could no longer be observed and the colour change takes place correctly within the game window.

see D4639

Arcanist wouldn't let me apply the patch, but doing it manually did the trick.
Patch fixes the problem described in the ticket.

a problem with openGL ARB was tracked down to this changeset.
see #6537

Apply the patch and compile the game

Ok

Compare performance for Combat Demo Huge

Before: ~170
After:~85
https://ttm.sh/bkP.jpg

see rP26852

Apply the patch and compile the game

1. Applied the patch with pull0ad.
2. When compiling, I get a myriad of new warnings. It did compile though.
   • Entire build log: https://ttm.sh/b4G.txt

Ok, tested on random/unknown, skirmishes/corinthian_isthmus_2p and skirmishes/forest_battle_4p maps. The issue from the ticket could not be reproduced.

An issues was found tracked down to this changeset. See #6531

see rP26849

In D4631#196937, @Langbart wrote:

Check that game looks like before (with different settings): water, shadows, particles, skies, trees, terrain, ao

(not yet, will upload some images later)

The mentioned performance difference between OpenGL and OpenGL ARB in D4631 is caused by this changeset. See ticket #6525 for more details.

bisecting leads me to this changeset for ticket #6524

Apply the patch and compile the game

It worked for the most part with arc patch D4631, but I had to create source/ps/containers/Span.h manually.
It did compile.

Check that game looks like before (with different settings): water, shadows, particles, skies, trees, terrain, ao

(not yet, will upload some images later)

Compare performance

• system_info.txt is here #5969
• I created a map with 5 players (me+4AI's) on a normal sized Ratumacos map. Entered the cheat code gift from the gods and have 91 horses run around the edge of the map..

Procedure (Visual Replay):

1. add profiler2.autoenable = true to user.cfg file
2. run the following command 0ad/binaries/system/pyrogenesis -mod=public -replay-visual=PATH
3. open profiler2.html in browser (but only when the loading screen disappears and the actual replay starts)
4. replay begins, choose paria and click Follow Player
5. replay ends, click Save Live Report to file
6. save profiler2.jsonp from the log folder and rename it accordingly
7. perform this for default settings with OpenGL and OpenGL ARB with/without applying the patch.

In D4623#196589, @sotirangelo wrote:

The time elapsed should display properly now.

Yes, it is fixed. I did not find any other issues when testing the patch. Seems to work as planned. Good job.

In rP26269#57641, @wraitii wrote:

Hello, thank you for taking the time to create a patch.
I noticed the following:

• Start a normal game e.g. Random/Guadalquivir River and press Shift+F5
• end the game, return to the Main menu and open Load game.
• The Played time is always 00:00

In rP26269#57636, @Stan wrote:

Does it work if you replace the ENSURE by an if ?

In rP26269#57626, @Stan wrote:

Can you be more specific about that issue ? Are we running into that Ensure ? And if so why ?

While working on #6513, we stumbled over this issue.

[01:42:34] Langbart could we do it like  wraitii  and make an if statement ?
[01:42:36] Langbart https://code.wildfiregames.com/D4459#change-tcKZz2mtyCmK
[01:44:26] elexis ehm
[01:45:21] elexis that was before bbs commit
[01:45:53] elexis if that texture manager commit is not reverted, then you possibly might not only may but have to add that if statement
...
[01:58:54] Langbart I am at rP26583 and the -autostart-nonvisual is running correctly.
[02:01:04] elexis did you test with a random map or skirmish/sceario?
[02:01:28] Langbart binaries/system/pyrogenesis -conf=mod.enabledmods:"public" -autostart="scenarios/arcadia" -autostart-nonvisual
[02:02:30] elexis this line
[02:02:30] elexis     ENSURE(CTerrainTextureManager::IsInitialised()); // we need this for the terrain properties (even when graphics are disabled)
[02:02:30] elexis totally looks like 26269 should have changed that
[02:03:59] elexis oh
[02:04:06] elexis try a random map
[02:04:13] elexis I tried the units demo map in that replay
[02:04:36] elexis that wasnt a random map actually, hm
[02:04:46] elexis a skirmish map that places the entities from a JS simulation script
[02:05:21] elexis so the question now is why does our code run CXMLReader::ReadTerrain but at rP26583 it doesnt
[02:08:53] elexis grep -R 'Terrain' -l | grep xml | sort
[02:09:00] elexis shows that only few maps have this element
[02:09:10] elexis units_demo.xml one of them
[02:09:22] elexis arcadia not
[02:09:41] elexis so try units_demo if you can
[02:10:22] elexis not posting nasty comments about sloppy commits
[02:11:47] Langbart elexis 
[02:11:50] Langbart u are correct
[02:11:55] Langbart arcadia works
[02:11:58] Langbart units dem does not
[02:12:05] Langbart i am at rP26583
[02:12:15] elexis so you can raise a concern there on that particular line
[02:12:23] Langbart do u want the lldb ?
[02:12:34] elexis it will be the same
[02:12:39] elexis the error is clear now
[02:12:48] elexis wraitii removed some lines but not all

update

In rP26815#57590, @vladislavbelov wrote:

Fixed in rP26833.

Ended up at this commit after bisecting.
#6520

lower cost pop to 3 and chnage specific name back to Tóushí Chē

IRC 27/Apr/22

[01:01:23] elexis and yes 26584 is conceptually broken by requiring GUI files while simultaneously requiring the GUI not to be initialized
[01:02:01] Langbart but it gave me "random" civs as an autostart option. this is great
[01:03:28] elexis I could argue this should have been considered in the (missing) use case analysis before r25077 already when gui/gamesettings/ was created. 26584 was to fix that hole already (bb made the patch after I ranted about that folder change). while the commits that introduced the redundancy of the gamesetup logic in Gamesetup.cpp gamesetup-js and Atlas gamesetup were to blame for this duplication of logic, it would have been warranted by the 
[01:03:29] elexis reaction of the folder and moving that much code (which comes at the penalty of code auditing cost if its perfect and in this case many (10+?) bugs)
[01:07:01] elexis I wonder if those slashes in "fix/ improve/ commit" mean "or" or "and"
[01:07:30] elexis void InitNonVisual(const CmdLineArgs& args) g_GUI = new CGUIManager(); just no 😃
[...]
[01:30:02] elexis D4527 would have to move  gui/gamesettings/ out of the gui folder and reflect on the best place

Thank you for quickly fixing the problem in the ticket with rP26829.

Some issues with particles were noted, wowgetoffmang mentioned that you can`t see the smoke on the Han forge and I have strange looking fires when buildings are destroyed.
#6519

"Human"

Ok

In D4610#196206, @vladislavbelov wrote:

Does the slowness happen immediately after recompile?

Every time I run recompile and start the replay the FPS is at 23 when the camels spawn (~20seconds) the FPS goes down to 9-11 and stays there for the rest, on the second run they reach 14-18 FPS.

What if don't run the replay for the first time, only open the main menu and close?

• Opening and immediately closing the game before starting the replay made no difference.
• If you wait a while for the Macbook to cool down between compiling and starting, it makes a difference. Could just be thermal throttling.

In D4610#196140, @vladislavbelov wrote:

@Langbart I've made few optimizations, could you check performance?

On the first run I started the replay immediately after compiling, maybe the game needed a test run before the performance test.
Now the performance felt more similar with and without the patch applied for openGL.

In D4610#196108, @Langbart wrote:

I personally felt that the performance was worse after applying the patch than without it, especially for openGL. It felt more choppy.

Apply the patch and compile the game

ok

Make sure the game looks like before the patch on GL and GL ARB

ok

Compare performance

• system_info.txt is here #5969
• I created a map with 5 players (me+4AI's) on a normal sized Lake map. Entered the cheat code gift from the gods and have 81 camels run around the edge of the map..

Procedure (Visual Replay):

1. add profiler2.autoenable = true to user.cfg file
2. run the following command 0ad/binaries/system/pyrogenesis -mod=public -replay-visual=PATH
3. open profiler2.html in browser (but only when the loading screen disappears and the actual replay starts)
4. replay begins, choose paria and click Follow Player
5. replay ends, click Save Live Report to file
6. save profiler2.jsonp from the log folder and rename it accordingly
7. perform this for default settings with OpenGL and OpenGL ARB with/without applying the patch.

In D4608#196071, @Freagarach wrote:

In D4608#196034, @Langbart wrote:

The unpacked siege units still have UnitMotion, but a very small almost-zero speed, and UnitAI doesn't allow them to move without packing anyway.

Do note this is wrong. I've noticed a long time ago that when you attack siege set to passive, interesting things happen to see by those who wait.

Adding UnitMotion

Build the game bundle and playtest it to see that nothing is broken.

The old cursor wins, the new one annoys me. wrong angle, wrong style, bad contrast

https

[14:55:22] elexis IMO on the screenshot it lacks the information what the user is supposed to do there, not everyone knows about the RGB color model
[14:55:28] elexis so it should be Red Green Blue IMO

These changes are not the cause of the problem from ticket #6486. They only made the problem visible.

In D4597#195732, @vladislavbelov wrote:

Could you attach screenshots?

Applying the changes from rP26780 to the GIT working copy under rP26777 fixes the problem described in ticket #6495 .

#6495

#6495

I don't see the connection between the 2 diffs, other than they both include towers. :)

The palisades_outpost template is removed, this diff replaces it with structures/brit/outpost and the other with structures/wooden_tower (structures/palisades_fort), both handle the issue from the ticket.

see D2740

A problem with a texture at death. #6494
Bisecting reports that this is the first commit in which the problem appears.

This solves the problem of the linked ticket.

@wowgetoffyourcellphone has already made a revision to the map in changeset rP26747.

Changing the camera angle was not intended.

In D4589#195396, @vladislavbelov wrote:

this solves the problem with the Viking Longship (ref #6486) and some objects were sitting too close to each other.

I believe it doesn't solve the issue but hides it. It seems if someone would make a test when the ship is pushed (or vice versa) the test might fail for the same reason.

Replace "MoveTo" with "JumpTo"; Remove unnecessary whitespace

This fixes the problem in the listed ticket.

turned looks better

@wowgetoffyourcellphone don't do the others with this angle, this one is bad enough. It obscures too much of the icon below. All the old ones were better.

#6486
Related to rP26680

D4585

#6486
I think you need a smaller depth number for your viking_longship.xml, at any number above 43 it breaks the Units Demo map.

#6484

Good commit. Siege are in good shape.

[art] The final new default cursor?

D4579