Engine Testing - 94.1.1-826-g7f6f61d (6. Aug 2013)

[jamerlan]

BTW looks like ETC2 texture compression is much better then S3TC. Maybe it's time to switch to it? http://habrastorage.org/storage2/a5f/c3 ... d21e35.png

This texture was included in OpenGL ES 3.0 and new Android users have it with android 4.3. I guess desktop users have this long time ago already. For example: http://www.phoronix.com/scan.php?page=n ... px=MTI0NTc

http://en.wikipedia.org/wiki/Ericsson_T ... ompression

I know that spring devs knows much more then me. But why S3TC still used?

[FLOZi]

Another crash on start http://paste.springfiles.com/view/0e7a1b8e

(actually start, not game load this time)

[Beherith]

Because dxtN has hardware support.

[jamerlan]

Because dxtN has hardware support.

As we can see - not in mesa by default. On other hand ETC is hardware accelerated even on android devices: http://software.intel.com/en-us/article ... ompression

[jK]

On other hand ETC is hardware accelerated even on android devices

This texture was included in OpenGL ES 3.0 and new Android users have it with android 4.3. I guess desktop users have this long time ago already. For example: http://www.phoronix.com/scan.php?page=n ... px=MTI0NTc

"This patch enables support for ETC2 compressed textures on all intel hardware. At present, ETC2 texture decoding is not available on intel hardware. So, compressed ETC2 texture data is decoded in software"

Some facts:
* there is only a ETC1 opnsource compressor available
* ETC2 is backward compatible with ETC1 (so spring uses ETC2, but the texture data is ETC1)
* ETC is only hw accelerated on smartphone gpus
* both Mesa-Intel & NVidia implement it in software atm
* duno if rest of Mesa supports it yet nor the state in ATI drivers
* so as long as drivers don't provide hw support it's not worth it
* Spring only runtime converts the terrain textures at runtime to ETC, unit texture are technically possible but not done yet
* so see it as a workaround of Mesa's white terrain issue, _just_ to improve user experience (most distros have S3TC disabled by default and don't install driver blobs).

[raaar]

I tested it a bit in single player vs AI

It seems to be working nicely.

but there's something strange with gunships:

1- order them to move in a direction
2- order them to move in the opposite direction while they are still moving

when order 2 is given, their bodies start turning but instead of decelerating immediately they keep going to direction of order 1 for a second or two until their body turns into the new direction. This kinda makes sense, but i don't think they did this before and they seem to do it in a weird way, they even seem to accelerate in the wrong direction if given the second order before having reached full speed.

[Google_Frog]

Gunship related ticket: http://springrts.com/mantis/view.php?id=3896

Apparently it is fixed (in that we can now configure it with turnInPlaceAngleLimit).

[PepeAmpere]

This bug still present: http://springrts.com/mantis/view.php?id=3937

Btw the new progress bar is nice and gives important info, what about to let game devs to customize it. At least positioning of game name and bar would be apreciated in future as good feature.

[smoth]

any possibility that we are moving to a new test version soon?

[FLOZi]

This bug still present: http://springrts.com/mantis/view.php?id=3937

Btw the new progress bar is nice and gives important info, what about to let game devs to customize it. At least positioning of game name and bar would be apreciated in future as good feature.

It is all drawn by lua, it is totally customisable already, simply change the code.

[PepeAmpere]

This bug still present: http://springrts.com/mantis/view.php?id=3937

Btw the new progress bar is nice and gives important info, what about to let game devs to customize it. At least positioning of game name and bar would be apreciated in future as good feature.

It is all drawn by lua, it is totally customisable already, simply change the code.

It was discussed here and on devs channel and noone had idea how to make unload on center spot, i think i was asking you, too, in december. But yes, its long time since we were talking about it.

the middle of the concrete tile is still not possible to be used for unload when i use Spring.SetUnitBlocking

But a bit away from center it can unload? imo report as new bug...

[Silentwings]

It is all drawn by lua, it is totally customisable already, simply change the code.

In theory, I've not seen any info on it... (and I've not tried either, presumably its just in basecontent?)

[FLOZi]

@Pepe:
I wasn't passing comment on 3937, but https://github.com/spring/spring/commit ... ff0f7adf82

@Silentwings:

https://github.com/spring/spring/tree/d ... tro/Addons

In particular;

https://github.com/spring/spring/blob/d ... ogress.lua

[Forboding Angel]

This bug still present: http://springrts.com/mantis/view.php?id=3937

Btw the new progress bar is nice and gives important info, what about to let game devs to customize it. At least positioning of game name and bar would be apreciated in future as good feature.

Load up evo test versions and be amazed.

[Jools]

The progress bar is very nice. But progress bars in general are a bit stupid, they like to get stuck on 99% or 100%. As I told in #moddev, we should make a little parody on most load indicators by making it get stuck on 99%.

Here's another parody on the same topic:

[zerver]

This is due to network latency or mechanical HDD seek time. With an SSD, the windows dialogs are fairly accurate, although very annoying due to the time wasted (to gather total file sizes and compute the time required for copying) before the actual copying starts.

[Jools]

But tried installing something like Photoshop? IIRC that get's stuck on 99% for about 5 minutes :)

[luckywaldo7]

The progress bar is very nice. But progress bars in general are a bit stupid, they like to get stuck on 99% or 100%. As I told in #moddev, we should make a little parody on most load indicators by making it get stuck on 99%.

It might be funny, once. But after a dozen games, it would definitely no longer be funny, and at 100, it's now the least funny joke ever, and after 500, it will be the cause for outright suicide in 4% of players.

So please, think of the players.

This bug still present: http://springrts.com/mantis/view.php?id=3937

Btw the new progress bar is nice and gives important info, what about to let game devs to customize it. At least positioning of game name and bar would be apreciated in future as good feature.

Load up evo test versions and be amazed.

Animated background and music? Nice!

[malric]

Is there any way to test performance automatically? (something like http://springrts.com/phpbb/viewtopic.ph ... e&start=40 but, reliable)

On XTA the new dev build seems as slow as the latest release (94.1) but without a clear number it is hard to understand if it is slower or slightly faster.

[gajop]

An issue from trying to test this (I prefer testing while working myself).

Setting
I've got a ZK installation in D:\games\Spring.
The root folder contains a couple of global settings, ZK Lobby and the datadirs: games, maps, etc.
It also contains an engine dir with the following structure:
85.
85.0
(etc.)
93.1
94.1
94.1-dev (folder i created by using pr-downloader as instructed by on the wiki)*

In the 94.1 folder I modified springsettings.cfg to have the following : SpringData = D:\games\Spring. That allows me to properly load any games/maps I would have normally been able to.

Problem
It seems that the root folder when running the 94.1-dev and 94.1 spring is different. In particular the root folder in 94.1 (and previous versions I think) is D:\games\Spring, while in the case of 94.1-dev it's D:\games\Spring\engine\94.1-dev.

This can be seen while running the following code:

Code: Select all

	x = io.open("exmp-file", "w")
	x:write("hello!")
	x:close()

Which on the 94.1 version creates the file in D:\games\Spring\, while in 94.1-dev it creates it in D:\games\Spring\engine\94.1-dev folder.

Question
Did I forget to configure something? Should I download this in a different way?

*To obtain this release I did the following:
1. Downloaded pr-downloader-0.7-12-g6dcf45f
2. cd-ed to the dir and ran: pr-downloader --download-engine 94.1.1-826-g7f6f61d
3. The previous command created files in My Documents\My Games\Spring\... so I copied the engine to D:\games\Spring\engine\, and renamed it to 94.1-dev