Version 2.4 sneaking in!

[nevincho]

I have e problem with new version. After run - crash.

last_run.log:

Attract-Mode v2.4.0 (Windows, SFML 2.5 +SWF +7z)
avcodec 58.18.100 / avformat 58.12.100 / swscale 5.1.100 / swresample 3.1.100

Config file not found: C:\MAME\pROGRAMS\Frontend\attract-v2.4.0-win32\attract.cfg
 - Constructed 1 filters in 0 ms (1 comparisons)

Any one?

[akafox]

In the next version I want it Attract Mode to pay off my debts...Okay Okay..I know let's be realistic..you can't make it do that.

But I DO expect to make my coffee for me in version 2.5 

No problems that I can see on Linux 64-bit smooth and awesome as always great job guys

[qqplayer]

Thanks for the update.
What do you mean exactly with "Attract-Mode is now looking for game overview information in a different place then previously" , I have tons of overview scrapped descriptions and I prefer to move them to another place than re-scrape all the stuff.
Which is the new location?

Here is the description of where the overviews are now stored:

https://github.com/mickelson/attract/commit/88595c811b2ebe717d1c9ea116373531b92cc6cb

hope that helps!

Thanks I have converted the "old ones" to the new "format".
Should be easy to make a python script if anyone wants to help.
I have used the find and replace feature of notepad ++ , just deleting the first "Overview " word at the begining and "n/n/"
Then just changing the .cfg extension for .txt with flexible renamer.

[zpaolo11x]

Some comments after playing a bit more with the new version:

- Issue #407 - fixed fe.Image.video_playing attribute

This was driving me nuts while trying to stop video playing in my layout, I tried all sort of tricks to avoid it and ultimately scrapped all the start/stop code So glad it's been fixed, maybe I'll have to recover the old code then...

- Added fe_file.cpp to load images, fonts, sounds through a file stream

What does this mean for a layout developer? Or is it only an internal change?

- Relabelled and reworked the 'minimum_wait_time' emulator setting

I'd like to have more info on this minimum wait time... anyone can help or provide a link?

For some reasons on Windows, when loading my theme 2.3 took some time to start, letting me wait in front of a white screen, now in 2.4 it starts much faster, like 2.3 on the Mac, which is very good.

Overall I managed to use the new zorder system, makes more sense to me.

I noticed my layout is a bit slower and a bit more stuttering on 2.4 with respect to 2.3, I made a side by side video since the difference is quite strong https://youtu.be/bXEFtRepDt4 . I used many tricks on my layout to make it smooth (rawset_index_offset, adding and updating hidden text objects in a nested surface), debugging this will be hard since this tricks I have no idea why they worked like they did lol

[zpaolo11x]

I think the reason why 2.4 feels slower on my machine is because of the way tick works now... it seems that the tick_callback ticks less often. To prove this I created a simple layout like this:

Code: [Select]

fe.add_ticks_callback( this, "tick" )

function tick(tick_time){
   print (tick_time + "\n")
}

And monitored the console output, for 2.4 it's like this:

63
103
118
133
148
163
178
193
196
211
226
229
244
259
262
277
292
295

while 2.3 is like this:

35
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168

as you can see 2.3 "ticks" almost every millisecond, while 2.4 doesn't. I noticed this because some tick-related scrolling routines were slower on 2.4, and i'm rewriting all these routines to work with timings. Any reason why this happens?

[Oomek]

From what I remember it was supposed to save the power on idle. The tick function is suppsed to be called on each redraw ( if any of the object’s properties has changed since the last frame ) not every millisecond.

[Oomek]

I noticed my layout is a bit slower and a bit more stuttering on 2.4 with respect to 2.3, I made a side by side video since the difference is quite strong https://youtu.be/bXEFtRepDt4 . I used many tricks on my layout to make it smooth (rawset_index_offset, adding and updating hidden text objects in a nested surface), debugging this will be hard since this tricks I have no idea why they worked like they did lol

add this to your layout:

Code: [Select]

local bar = fe.add_text( "", 0, 0, 32, flh )
bar.set_bg_rgb( 255, 255, 255 )
bar.bg_alpha = 200


function tick( ttime ) {
bar.x += 4
if ( bar.x > 1920 ) bar.x = 0
}
fe.add_ticks_callback( "tick" )

and start disabling objects. If you'll manage to identify the cause of stuttering I can look and see why it happens.

[zpaolo11x]

From what I remember it was supposed to save the power on idle. The tick function is suppsed to be called on each redraw ( if any of the object’s properties has changed since the last frame ) not every millisecond.

Understood, the millisecond timing is probably because I did the test on a Xeon workstation that has plenty of power... I'll check with the code you sent me after the holidays. Looking at the numbers it seems that 2.4 is ticking at roughly 60Hz, which makes sense, while 2.3 was ticking way too fast. I guess then that relying on a counter in a tick_callback is not a good way to manage transitions

[iOtero]

In Windows and in Macos High Sierra it works very well, but it does not work in mac os x Mountain Lion... 

Would there be some way to make version 2.4 work in Mountain Lion? Version 2.2 of AM was compatible. 

[Oomek]

Could you run AM  with —loglevel debug parameter and post a log here? (It’s double dash - - without a space)

[iOtero]

Could you run AM  with —loglevel debug parameter and post a log here? (It’s double dash - - without a space)

In Windows, this command

Code: [Select]

attract --logleveldoes not work in AM 2.4 (not generate "last_run.log").

Yes it does in AM 2.3

If i run AM 2.4 without the "--loglevel" option, AM does generate the "last_run.log" file. 

In MAC, with AM 2.3 and AM 2.4, in terminal, "open Attract.app" not run. Shows "LSOpenURLsWithRole() failed with error -1810 for the file /Applications/Attract.app."
And running attract from the launchpad. It does nothing, nor generates any log, that i have seen. And i've searched everywhere for the "last_run.log" without finding it.
AM 2.2 runs perfectly, but it does not work "per display" option. 

[Oomek]

It’s

Code: [Select]

attract --loglevel debug

[iOtero]

About Attrac-Mode 2.4 at Mountain Lion, nothing can be done.

Version 2.4 searches for libraries that Mountain Lion does not have and that can only be installed through homebrew.

And in Mountain Lion now there is no way to install homebrew.

So you have to settle for version 2.2 

[keilmillerjr]

About Attrac-Mode 2.4 at Mountain Lion, nothing can be done.

Version 2.4 searches for libraries that Mountain Lion does not have and that can only be installed through homebrew.

And in Mountain Lion now there is no way to install homebrew.

So you have to settle for version 2.2 

False. I had it installed on mountain lion. You need to install XQuartz I believe. I’ve been consistently running brew for a long time now.

[nevincho]

New version can`t create .cfg file

last_run.log:

Attract-Mode v2.4.0 (Windows, SFML 2.5 +SWF +7z)
avcodec 58.18.100 / avformat 58.12.100 / swscale 5.1.100 / swresample 3.1.100

Config file not found: C:\MAME\pROGRAMS\Frontend\attract-v2.4.0-win32\attract.cfg
 - Constructed 1 filters in 0 ms (1 comparisons)

Pls help.