Ren'Py 6.99 Prereleases

[nyaatrap]

Does drag&drop still update in real time? To me, draggables jump to a drop position without moving.

[PyTom]

It should, but the first prerelease had a huge number of performance issues that the next prerelease should deal with.

[PyTom]

6.99.0.244)

Sorry about the long delay between prereleases. Apart from some family medical problems that slowed me down, the first prerelease proved to have a lot of hard problems, and I wanted to fix them all before I released again. I think I managed to fix all the problems reported in this thread, as well as those reported at github and elsewhere, but for some of the problems I had to make an educated guess as to what to fixed.

The changes include:

Updated Spanish and Japanese translations.

Ren'Py now uses the integrated GPU on macs by default, to save power on laptops.

PNG writing has been fixed.

A number of drag and drop issues have been fixed, and the new drag_offscreen property has been added to allow drags to be dragged partially offscreen.

The ATL contains statement has been improved - it now copies transforms when showing them. (Similar to the way the show statement does.) An issue has been fixes where ATL would run a transform during prediction, and "lock in" the values at that time - even if the values changed between prediction and show. That's been fixed.

The onscreen keyboard is now re-opened on Android if the user closes it during an input statement.

We use FBOs to do render-to-texture on Android. Hopefully, this will fix the Dissolve problem on Android lollipop that people have been reporting. (Please let me know if this fixes the problem.)

The renpy.copy_save function has been added.

Ren'Py now corrects absurd file timestamps when creating a zipfile.

A Space Whale-sized memory leak has been fixed. (A self-reference was preventing surfaces from being garbage collected by the reference-counting collector, but the memory in those surfaces was being allocated by SDL, not Python, so the Python cyclic GC wouldn't run.)

Skipping will no longer prevent Ren'Py from processing keyboard events (like the key to stop skipping).

Ren'Py will now detect infinite loops and report them as an exception. (Only in Ren'Py code.)

The longstanding imagemap cache problem has been fixed.


You can get 6.99 by switching to the prerelease channel and updating, or visiting http://www.renpy.org/release/6.99 . Thanks to everyone who reported bugs - your work made Ren'Py much better, by finding things I couldn't. And thanks in advance to everyone who reports bugs in .244 - hopefully fewer this time.

[crimsonnight]

6.99.0.244)

Sorry about the long delay between prereleases. Apart from some family medical problems that slowed me down, the first prerelease proved to have a lot of hard problems, and I wanted to fix them all before I released again. I think I managed to fix all the problems reported in this thread, as well as those reported at github and elsewhere, but for some of the problems I had to make an educated guess as to what to fixed.

The changes include:

Updated Spanish and Japanese translations.

Ren'Py now uses the integrated GPU on macs by default, to save power on laptops.

PNG writing has been fixed.

A number of drag and drop issues have been fixed, and the new drag_offscreen property has been added to allow drags to be dragged partially offscreen.

The ATL contains statement has been improved - it now copies transforms when showing them. (Similar to the way the show statement does.) An issue has been fixes where ATL would run a transform during prediction, and "lock in" the values at that time - even if the values changed between prediction and show. That's been fixed.

The onscreen keyboard is now re-opened on Android if the user closes it during an input statement.

We use FBOs to do render-to-texture on Android. Hopefully, this will fix the Dissolve problem on Android lollipop that people have been reporting. (Please let me know if this fixes the problem.)

The renpy.copy_save function has been added.

Ren'Py now corrects absurd file timestamps when creating a zipfile.

A Space Whale-sized memory leak has been fixed. (A self-reference was preventing surfaces from being garbage collected by the reference-counting collector, but the memory in those surfaces was being allocated by SDL, not Python, so the Python cyclic GC wouldn't run.)

Skipping will no longer prevent Ren'Py from processing keyboard events (like the key to stop skipping).

Ren'Py will now detect infinite loops and report them as an exception. (Only in Ren'Py code.)

The longstanding imagemap cache problem has been fixed.


You can get 6.99 by switching to the prerelease channel and updating, or visiting http://www.renpy.org/release/6.99 . Thanks to everyone who reported bugs - your work made Ren'Py much better, by finding things I couldn't. And thanks in advance to everyone who reports bugs in .244 - hopefully fewer this time.

Thanks Tom, I'll be able to check if the transition issue on Lollipop has been fixed later

[jack_norton]

Unfortunately the crash I have at end of any game (or exiting launcher) is still there even in the latest version

[PyTom]

Can you send me one of the games built with the new release? On what platform are you getting the crash?

[nyaatrap]

The drag&drop issue is still possessed on me. I don't know what triggers this behavior on my script. I'll look up on it later (I tried several thing but no clue now. It's on dorepara which I released. In this game, after hiding the window allows dragging a sprite).

[crimsonnight]

The screen flipping issue on Lollipop has been solved as well as the keyboard issue, so it seems there are no longer any big issues on android, great work!

[PyTom]

The screen flipping issue on Lollipop has been solved as well as the keyboard issue, so it seems there are no longer any big issues on android, great work!

Thanks! Reports of fixed bugs and successes are very useful.

The drag&drop issue is still possessed on me. I don't know what triggers this behavior on my script. I'll look up on it later (I tried several thing but no clue now. It's on dorepara which I released. In this game, after hiding the window allows dragging a sprite).

I was able to repeat this, so I'll look into it soon.

[Samu-kun]

Good news, the memory leak appears to be fixed.

nyaatrap's drag&drop bug also happens in Sunrider Academy.

Also, I've noticed that holding ctrl now scrolls the text much slower... Maybe about 20% of the previous speed.

[jack_norton]

Just thought to let you know that my issue was 99% due to video drivers, I downgraded to some older one (but more stable) and the problem went away! Damn ATi cards...

[Dharker]

Hi,

Just wondered if anyone has tested the steam achievements yet? I would like to test the achievements on my sword of asumi game, but following the documentation I figured once the steam achievement is setup (named: intro_killing) then I would simply need to input:

Code: Select all

    if people_killed == 2:
        achievement.grant(intro_killing)

Yet adding this in pulls an error on startup:

File "game/prologue.rpy", line 132: expected statement.
achievement.grant(intro_killing)

I'm probably being stupid, please enlighten me if i am.

[PyTom]

A working version of that code is:

Code: Select all

    if people_killed == 2:
        $ achievement.grant("intro_killing")

Note that I haven't released the steam files yet, due to busyness, so this won't sync to steam quite yet.

[Dharker]

Thanks for working version, its great to see I was just being stupid, and adding steam achievements will be easy

Sorry for pre-empting the steam part, I thought they were included in this pre-release. I will eagerly await those being added so i can test the syncing with steam.

[Zetsubou]

Renpy 6.99.0.244
Xubuntu 14.04 (xfwm 4.12, also tested with xfwm 4.11)

When I swap workspaces, the Renpy launcher and any games I'm running will follow me to the new workspace.
This did not happen in 6.18 (haven't tested on any other 6.99 pre-releases), and toggling workspace visibility or restarting the Renpy launcher doesn't help.
I'm guessing this is an SDL2 issue, but surely the default behavior shouldn't be to always be on the active workspace?