In D2531#108430, @Stan wrote:

In D2531#108428, @vladislavbelov wrote:

I've reproduced that assertion locally on macOS 10.14.6 in debug-mode. I've figured out the reason: it seems that OSAtomicAdd64 doesn't work as expected in debug mode, maybe address was misaligned.

Can it be fixed?

I've reproduced that assertion locally on macOS 10.14.6 in debug-mode. I've figured out the reason: it seems that OSAtomicAdd64 doesn't work as expected in debug mode, maybe address was misaligned.

Refs D2476.

MemorySizeMiB and its usage was added in rP9546, was removed in rP13892.

In rP23423#41006, @Nescio wrote:

This is a great improvement, many thanks!

In D2567#108291, @Stan wrote:

Accepting so you can close.

Fixes @Stan notes.

Looks ok for me.

I've tested the patch on Windows VS2015, it works and looks good for me.

I've reviewed and tested the patch. It looks ok for me. I'm going to commit it soon (fixed years and GL types by myself).

In D2545#107432, @bb wrote:

D2546 would require to change this

The rest of the patch looks good for me.

Used CStr8 instead of std::string.

Adds division by zero test.

In D2574#107740, @Stan wrote:

Maybe add a test for division by 0 and #elif?

Did you use master version of release version (https://github.com/OGRECave/ogre/releases/tag/v1.12.4)?

Years. Also it's important to have a test for the new terrain functionality. But I suggest to split the patch and change the CTerrain and tests in a separate patch.

Refs D2567.

In D2542#107484, @Stan wrote:

See the macos warnings https://jenkins.wildfiregames.com/blue/organizations/jenkins/macos-differential/detail/macos-differential/117/pipeline

In rP23387#40848, @Nescio wrote:

Thanks!
Any ideas on how to remove the unnecessary empty space between the logo and the “Learn To Play” button (red)?

@Nescio I've tweaked the algorithm. Now it's a bit closer to your understanding. Now words separated by any non-letter characters should present in a name.

Tweaks the algorithm and adds an immediate reaction to the checkbox.

Wrong ref, right ref is D2476.

Fixes notes.

In D2416#106689, @Angen wrote:

Do you know why that would be usefull?

Nope, just wondering.

I'm going to commit removing of the system cursor soon.

In D2557#106791, @Itms wrote:

That's true. I don't think we have other occurrences of in-game terrain changes though (especially since, before this patch, they would be quite expensive).

In D2557#106783, @Itms wrote:

An extra test (which I didn't run but would be needed) is a serialization test on a map where the terrain changes during the game: typically Extinct Volcano.

What was in case !m_BoneMatrices && !anim->m_AnimDef before the patch? It seems it returns true.

In D2537#106528, @nani wrote:

Adding a extra optional sharpening pass in fxaa.xml should improve the output image.
I propose this simple box sharpening one that I use in my mod that improves the image quality.

I think it'd be useful to attach a side-by-side comparison image.

In D2542#106381, @Nescio wrote:

A bit unrelated to the patch, but I see the problem with the quarter of screen. Could you provide the system information (system_info.txt might be good)? Also do you have a system scale?

In D2542#106381, @Nescio wrote:

The old function returns too little (no matches) for my taste and the new one too much.

I can reorder it and have more useful items first.

In D2542#106377, @Nescio wrote:

healer_ returns healer_b, healer_a, healer_e times 13 civs in the old function, i.e. 39 units; in the new function the first item is units/kush_support_healer_a, but units/athen_infantry_slinger_b is number seven and skirmish/structures/default_sentry_tower is listed above units/athen_support_healer_a; the item listed last is units/viking_longship.
With sublime I assume you mean some specific software I don't know.
What's intuitive varies from person to person. What I'd prefer is a search function that's somewhere between the old exact single string match and the new single character match.

In D2542#106375, @Stan wrote:

Would be nice to have regexp support?

In D2542#106371, @Nescio wrote:

That's a bit unexpected.

That might happen.

In D2542#106367, @Nescio wrote:

So it will give an ordered list of everything that has at least a single matching character?

In D2542#106351, @Nescio wrote:

So how does the new search work then?

In D2542#106334, @Nescio wrote:

But then I don't understand why you gave units ptol hlr a and not unitsptolhlra or ahillnoprsttu. Besides, there are no entities or actors with a space in their file name.

In D2542#106329, @Nescio wrote:

Isn't the space a separator that functions as AND? To me hlr means hlr but h l r means h and l and r. At least that's how search works in programs such as Gnome Nautilus/Files:

In D2542#106321, @Nescio wrote:

Then perhaps make it the default and have an “exact name” checkbox?

In D2542#106315, @Nescio wrote:

Toggle the checkbox "Advanced search". Maybe it has misleading name.

Thanks. Toggling the checkbox and then entering the string works. Entering the string first and then clicking the checkbox does not.
Why is there a checkbox anyway? Why not make it the default?

In D2542#106293, @Nescio wrote:

It seems Atlas fails to find anything when a space is entered in the search field, as was already the case without this patch. Any ideas what I did wrong?

Toggle the checkbox "Advanced search". Maybe it has misleading name.