You can now close this revision, which was committed in rP27806 partially.

I think the Fcollada fix is okay, we could ask the guy maintaining the rdb fork what he thinks about it.
If @vladislavbelov thinks it's okay too I have no objections.

D4085

This seems to have broken the Windows CI.

D2432 D4283

Would it make sense to use the Status enum ? e.g INFO::OK, INFO:SKIPPED ERR:* ?

<insert the macro for x86 detection> that would be incorrect. Some other arches support it. I don't remember but I think one of power pc xx or elbrus e2k support it.

It feels weird for it to be an option as it makes no sense to even be able to enable it on aarch64, arm64 or arm.
ma
https://github.com/0ad/0ad/blob/master/source/lib/sysdep/compiler.h#L106 makes me wonder what happens when building with clang. The CI doesn't print warnings about using the slow codepath (unlike arm)

In D4882#214666, @sera wrote:

COMPILER_HAS_SSE is a pretty bad name given it's description, we don't care at all if the compiler supports SSE. What we care about is if the target we compile for has SSE. With target I don't mean specific hardware but if -msse2 or -mno-sse2 is specified when supported respectively their equivalents for various compilers.

For provided binaries, Mac and Windows, offering only SSE enabled ones is fine, so runtime check can be removed in my eyes.

I'd use #ifdef SSEn and add those desired defines to premake for Windows, or at least move it to config.h, compiler.h feels wrong in any case.

In D4882#214662, @vladislavbelov wrote:

In D4882#214661, @Stan wrote:

The plan was to switch on runtime depending on the machine.

Doesn't SM require to use SSE?

Btw I'm not sure the description is accurate. We do not require it on the client side, but on the building side. The plan was to switch on runtime depending on the machine.

cc @Freagarach @wraitii

In both cases should be in the wiki. Also I believe some occurences have already been committed without that extra space.
EDIT: Another question is can it be enforced by say, clang format

Brace or space ?

You can undo the translation change if you need to.

Up.

I fiddle around but there was no way to stop the map from uncovering

@phosit could it be what you noticed in D4786?

@Dunedan you need to close this since you forgot to add Differential Revision: https://code.wildfiregames.com/D4999 to the commit message.

Maybe it should be a game option instead? 300 is likely to be more problematic than 200 on a lot of hardware

One thing to keep in mind is translators have a hard time translating everything that is not straightforward.

For my knowledge what's the advantage of a struct rather than a lambda ?

Can you add one in the soundmanager too?

I'm not sure it's very readable most people don't even know wha a^b is. Also it's divided not devided.

Seems to fix the main menu errors.

On a clean SVN after applying this patch I can no longer launch the game through the vulkan SDK. This does not happen without the patch. Running it without works, but no warnings...

Yeah they have a commit that says "indent stuff"

cc @Mercury @jprahman

Can you update the version of build.sh so it actually builds it on the CI ?

What happens if the value is NaN/null?

You can mark this diff as planned changes in this case.

Just a thought I had, what happens if you use Direct Connection instead of going through the lobby, with a fake username.
Maybe it would be better to use the actual jid?

Wonder if PLAYER_GAIA should have a constant.

Ah nevermind, finally got what you meant. I tested, could not break it, tried AIs, 8 players + 1 observer, moving units around, changing perspective. Also atlas

Doesn't pass the debug build on Windows...

Just in case you don't connect to IRC: This does not compile so I'm pretty sure I can't test it much :P

You're missing one change in premake5.lua See: https://code.wildfiregames.com/rP27440

No objection on my end.

Ticket about commands: #6804

Alright then. Let's hope for the best. D4775 is something similar although it only prevents sending useless stuff to one of the clients.

Well so far I haven't seen any player compile the game, install custom mods though, more often.

@animus It would be nice for the other two to be consistent as well (would get rid of those linter warnings), can you update the patch?

Notice this diff also removes the chat leak (as in you can no longer read your opponents chat messages using the mainlog.html) it's not just about removing IPs.

Fix macro

Less suppression. There is almost nothing in the logs anymore anyway since logging messages leaks chat.
But this is a patch I'm forced to make rather than I want to make. So if anyone feels like taking over please do so.

Actually we're not logging for a reason :D The only logs that I know are useful are the ones we're hiding. I doubt anyone ever uses the turn numbers for anything other than debugging, which is what a debug build is for :P

@Itms any opinion on this?

Try to fix compilation, reduce duplication.

@Silier, Ideally I wouldn't have to disable anything... Are they useful for developpers or useful for players?

Yeah not this alpha