The Settlement.js class should be updated/deleted/renamed :)

Segfault in tests for the record

I guess unit carrying resources or on a resource should obey orders and leave the wall or maybe a flag in the garrison holder that says unit are able to 'dismount' and should ignore orders

In D2377#99209, @smiley wrote:

Since apparently I had commented on this before, I have a feeling this would not quite work the way it should in certain cases.
Then again, its just a feeling. Consider that before wasting hours on potentially nothing.

Original patch was made by Amplikon, but has been mostly rewritten.

That inline got me wondering, wouldn't a saner implementation be to specify the build animation and fallback in the cmpIdentity ?

Can confirm it also affects windows + 10% CPU

I checked only checkrefs.pl warnings. Death and Sitting should be reexported

Remove more duplication.

Fix some comments

In D1190#98997, @Imarok wrote:

Quickly tested the patch. Looks like a really nice behaviour. 👍
But here are some things I noticed:

• I think elexis mentioned it already: There is no way to get rid of the ghosts of dead towers.

Remove warning,
Make green turrets disappear when cancelling build

Rebase. Fix a few comments.

See IRC today, where the current configuration already allows for options, as one can press ALT+C or ALT+V to disable some overlays when in TAB mode.

Thanks a lot for working on this :)

@OptimusShepard will you update the patch ? :)

Fix ifndef now that the file got back its name.

Include fix.

Rename Wrapper to fit the classes in that folder.

Move declaration to constructor

• Add setup function
• use std::string and c_str()

Try to fix tests

Diff with original version

• Replace std::string by CStr8 (We need the length because the code casts Token to const char)
• Move * pointer to the left
• Move everything to graphics
• Move tests next to the new code

A few bugs, changing ownership in atlas (numpad keys) kills the turret instead of changing its ownership (Might be were the problem comes from)
Garrisoning a new unit doesn't put in a visible spot.
You edited the brit champion chariot but forgot to add a unit on top :)

Regarding static_cast<off_t>(expr): I admit I never got around with the various kinds of casts offered by C++ (when I design SW on my own, I try and avoid casting alltogether). When updating the patch, I can still put it in.

On the usage of std::set

It doesn't like your testing code, only works properly on mingw32 https://code.wildfiregames.com/P185

Most of the time everything is reset, because you change the flag position completely. The only case where you don't reset the vector is when you shift click IIRC.

Just tested in Atlas and Combat Huge Demo without seeing any issues. Also run the tests successfully.

In D1682#72623, @Stan wrote:

If you want more specific output https://pastebin.com/myEAnKEj
9000000 samples
cl.exe fine.cpp /EHsc
STD: Vector 'push_back' time:1.62
STD: Deque 'push_back' time: 2.193
STD: Vector 'emplace_back' time: 1.551
STD: Deque 'emplace_back' time: 2.222
STD: Vector 'erase' time: 1.236
STD: Deque 'erase' time: 10.948
cl.exe fine.cpp /EHsc /O1
STD: Vector 'push_back' time:0.428
STD: Deque 'push_back' time: 0.771
STD: Vector 'emplace_back' time: 0.468
STD: Deque 'emplace_back' time: 0.689
STD: Vector 'erase' time: 0.074
STD: Deque 'erase' time: 1.025
cl.exe fine.cpp /EHsc /O2
STD: Vector 'push_back' time:0.297
STD: Deque 'push_back' time: 0.647
STD: Vector 'emplace_back' time: 0.341
STD: Deque 'emplace_back' time: 0.64
STD: Vector 'erase' time: 0.029
STD: Deque 'erase' time: 0.62
cl.exe fine.cpp /EHsc /OX
STD: Vector 'push_back' time:0.299
STD: Deque 'push_back' time: 0.669
STD: Vector 'emplace_back' time: 0.345
STD: Deque 'emplace_back' time: 0.641
STD: Vector 'erase' time: 0.027
STD: Deque 'erase' time: 0.628

It sounds weird to do that considering some of the names in the civ files already contains numbers ?

I had the same issue for D2338. Unfortunately the only solution I found was to clone the git repo, and do it there.

Update copyright. @vladislavbelov I think the license bear does strange things.

• ogre3d → ogre3d_preprocessor.
• Try to fix build.

We must note that the error callback was removed in https://github.com/OGRECave/ogre/commit/856c276ebd4cb7100f4b133b0e5b29cbc8558653

Could restart the build, was broken because of checkrefs.pl.

Indeed debug_printf doesn't work on windows but that's because we use OutputDebugStringW which redirects to a special debug output instead https://stackoverflow.com/questions/55432747/showing-outputdebugstring-messages-with-window-application

I guess @Itms could add it to his script when he has some time :)

Update headers.

Now in git format from a git repo ladies and gentlemen.

• Adds rotation to the mix, showing that no information is there about the negative Z
• Hopefully fix the FromJsval() I'm not really confident with those things.

Maybe this could be of interest https://stackoverflow.com/questions/4899657/performance-impact-of-virtual-inheritance

Also I'm not sure we can Use sproppoint as it does not contain the correct prop location and will always need to be modified anyway

I assume this sentence is written because Phabricator refuses empty test fields.

Actually no, I guess I could improve it a bit. Else I would just have written a blank space.

• Rename VisibleGarrisonAllowedClasses to AllowedClasses
• Fix some linters warnings

No I did not I'll see if that bring back the minus. Still lies the issue of non visual...

Remove all changes and just run svn diff -x -U5000 > file.diff

Try to fix the diff

Fixed by rP23016

Cause I tested it and that was the bug reported ?

Yeah was a trade cart.