similar issue solved in rP25245

replace op="mul" with op="mul_round" ?

In rP27241#59224, @Freagarach wrote:

#6659, but I'm a bit lost why,,, It works fine for mercenaries.

after applying the patch, the problem from the ticket is no longer reproducible. The issue is fixed .

P238

D4732

P277 complains as well. (There should be more template files listed, but after the first entry it stops.)

❯ python3 Production.py -v -r .
INFO - Launching 0.A.D with the following command binaries/system/pyrogenesis -mod="mod" -mod="public" --rl-interface="127.0.0.1:9090" --autostart-nonvisual --autostart="skirmishes/acropolis_bay_2p"
INFO - Calling http://127.0.0.1:9090/evaluate to get data
INFO - Killing 0 A.D.
INFO - Looking for templates with missing production queue.
ERROR - structures/generic_field has no production queue.

• The execution of the script from the public folder worked.
  • Testing if comments will be removed from the .xml file.
• The comment was still there after running the script.
• The only negative side effect is that the comment did not move with the renamed component.

• Try to start the sandbox map from the Han, many errors.
• Bisecting lead to this commit.

Only bad side effect is the comments are thrown under the bus.

i was missing lxml

• BeautifulSoup#installing-a-parser

I could not get it to work.

• asciinema.org

Restore catalog.xml

It fixes all the problems referenced in the ticket.

In D4511#194122, @Langbart wrote:

Maybe even more vision, I can't see the progress on bombing the house unless I move closer.

make a dangling symlink ...

❯ ln -s ~/ThisFolderDoesNotExist ~/Library/Application\ Support/0ad/mods/ThisFolderDoesNotExist
❯ ls -la ~/Library/Application\ Support/0ad/mods/ | grep ThisFolderDoesNotExist
lrwxr-xr-x   1 paria  staff     35 Jul 24 22:55 ThisFolderDoesNotExist -> /Users/paria/ThisFolderDoesNotExist

start 0ad

Rebase

rebase

Mostly arc patch gets the job done, if not I use one of @wraitii's formulae:

• pull0ad or
• P246.

However, they also do not filter out empty files.

In D4735#201575, @lyv wrote:

the files are always empty and not deleted

If you are applying with patch, you can use the -E flag. I generate patches using git diff which has a slightly different format (even phab shows those files as just deleted while this revision shows the files cleared). If you are using arc patch I have no idea what to do.

In D4735#201530, @marder wrote:

I think the problem is I don't use arcanist to upload the patch to phabricator since it is stupidly annoying to setup on windows.

😑

• vlad is on windows also does not use arcanist and on his patches it works. (D4434)
• works for smiley too (D4611)

In D4735#201530, @marder wrote:

delete the sele & mace library templates

You need to find out how to delete these files correctly when you create a patch. Every time I try to apply one of your patches with deleted files, the files are always empty and not deleted, I have to do it manually. This is also the reason why the ci in D4713 failed.

#6587
s0600204 wrote:

The issue with the Structure Tree is that r26015 removed default Researcher/TechCostMultiplier values; and whilst r26192 fixed that for the simulation, a fix was not applied to globalscripts which expects there to be values for each possible resource specified within templates.
This is causing other technology costs to be displayed incorrectly within the Structure Tree, such as those of the Mauryan temple and the Gallic forge

rP27009

General

• The fortress should come after the defense_tower

Civ list

athen

• @wraitii has already began something differnt here, see rP25239
• that is why the gymnasium and prytaneion is listed before the wonder

• super_dock should sit next to the wonder (only accessible in phase 3)

• they have one extra building now

• they have one extra building now

#6593

With rP26996 there are two new techs phase_city_pers.json and phase_town_pers.json, both are (wrongly) listed by the checkrefs.py script, see log output below for more clarity.

unused aura.
#6544

I am happy with it, it fixes all the errors from the ticket.

It found all unused files from the current change set and 26899. I also tested if the name of an aura/tech was the same but in a different folder, it found those correctly too. It even found techs where the autoResearch was set to false. 👌

After applying the patch the python script successfully found the unused han_hero_wei_qing_3.json file.

Sorry for not being clearer!

It fixes the first issue in the referenced ticket, the second issue, the selection of units, can still be reproduced after applying the patch.
Will attach a replay with latest GIT version + D4727.

I have to say that I am very disappointed with the decision to proceed with this patch. Detailed comments were simply ignored I will definitely not continue to work on this project if this carries on.

rP26961

• reduce the size of the dialog for the openURL(url) function, it was added with rP15062 and is with 600px unnecessarily long
  • The box showing the link is very useless and can be removed, it is broken anyway for very long links e.g. Multiplayer → Game Lobby → Terms of Service → View online
    • The conformation box for opening URL's has already be removed for Mod selection → Help → Modding Guide/ Visit mod.io
• reduce the button size for the case there is only one button on the dialog box
  • make it adoptive to the text length a little bit, but the button should not stretch across the entire message box window
• Pick a height for the red buttons, just enough to hold two lines of text and the proportions of the message box are aesthetically pleasing, 40px looks good. These confirmation options are not supposed to be very long. The string "Quit and View Summary" from rP25978 should have simply be "Summary".
  • Alternatively

my idea would be a two size solution. One for short words / one lines and one for two lines.

If you have too much time, wrap the function in a loop and see if the probability of each civ is now roughly equal when you draw 1000000 times.

Setup:

• create LongStringLocale for bundled languages (build-archives.sh).

python3 generateDebugTranslation.py --long --languages "ast" "ca" "cs" "de" "el" "en_GB" "es" "eu" "fi" "fr" "gd" "hu" "id" "it" "nl" "pl" "pt_BR" "ru" "sk" "sv" "tr" "uk"

• start

binaries/system/pyrogenesis -conf=locale:long

Impact

rP26969

#6578

rP26950

Apply the patch and compile the game

ok

Check big maps like African Planes or Egypt 3v3

tested Egypt 3v3

Compare performance via F11 (especially gui rendering)

Run entvalidate.py

Apply the patch and compile the game

ok

Follow the steps from #6569

Tested with delenda and with replays from the actual game for openGL & ARB. The problem from the ticket could not be reproduced. The line of sight updates accordingly when moving your units, constructing structures or doing tech vision upgrades. Checking the minimap, testing different shadow options, switching between civilisations and different game speeds could not reveal any problems.

Ternary it shall be.

#6566

#6569

#6560

rP26933

rP26933

discuss the strings that should be displayed

better VICTORIOUS and DEFEATED
see summary tooltip

Music.prototype.DEFAULT_VICTORY_TRACKS = [
	"You_are_Victorious!.ogg"
];

and for defeated YOU HAVE BEEN DEFEATED!

For me it compiles and runs. IDK were the code is used.

Yes, for me too. macOS10.15

This problem can still be seen on the Mainland map with the generic/sahara biome for the Han civic centre.

With the latest update of this patch, I can no longer reproduce the problem described in the ticket.

Check LOS changing by moving unit with enabled/disabled frame limitation with GL and GL ARB

Just some videos, they both look fine.

Apply the patch and compile the game

ok

Check LOS changing by moving unit with enabled/disabled frame limitation with GL and GL ARB

See patch D4681.
This commit makes the problem more visible, but is not the actual cause of the bug.

In D4677#199006, @Stan wrote:

@Langbart can you try to build this on your mac? It seems the tests fail to build.

The stone mine on the Anatolian Plateau looks better with this patch.

It fixes the problem described in the ticket.

#6552

#6553

#6553

#6546

With the default settings, it now looks more elegant and beautiful.

Apply the patch and compile the game

ok

Check that minimap looks like before with GL and GL ARB backends

ok

Check that map revealing looks like before
Check that LOS behaves like before

Yes, but it should be noted as soon as I added smoothlos = "true" to my user.cfg without the patch being applied I would crash for any rendererbackend. After applying the patch I could set smoothlos to "true" and would not crash for both rendererbackends.

Patch applied

Units demo

openGL: https://ttm.sh/bad.43.png
openGL ARB: https://ttm.sh/ba2.01.png

elephant_worker.json could be removed.

Apply the patch and compile the game

Ok

Check that GL minimap looks similar/same to GL ARB

Minimap looks good, but look at the textures on the scenarios/arcadia map. They appear more squared on openGL than on openGL ARB.