• Hello [name]! Thanks for joining the GMC. Before making any posts in the Tech Support forum, can we suggest you read the forum rules? These are simple guidelines that we ask you to follow so that you can get the best help possible for your issue.

OFFICIAL GMS2 Version 2.1.5

I had a an extension added to my game that no longer works. Its a "virtual keyboard" extension. It's odd, because it works perfectly for PC but when I want to use it in an android release, it doesn't work properly. It does not show the string written, but rather just shows an empty string. It's odd because it doesn't work on android, I don't know if it has anything to do with the new features of virtual keyboard being added. Other than that, I would really like to know how these virual keyboard functions work. I got the keyboard to show, but don't know how to close it when I press the "done" return key.
 
E

Excalipur

Guest
With the new runtime 2.1.5.246 when I try to ryn my game both in VM/YYC compiler crashes and I get an AssetCompiler error message box which says "Index was outside the bounds of the array". I switched back to runtime 2.1.4.218 and everything runs smoothly there.
I have this exact same problem... It doesn't happen with all my projects. The compiler gets to "Writing chunk... SOND." And then I get the super non-descript crash you mention above. Going back to an older version of the project does not solve the problem which means it's definitely caused by the update.

EDIT: Reverting to the previous run-time fixes it for now, but sadly bars me from the new functions.
 
Last edited by a moderator:

sman

Member
When I click the Run button for Android, nothing happens, not even in the Output window... :(
Reverting to the previous run-time changes nothing.
 
Last edited:

rmanthorp

GameMaker Staff
Admin
GameMaker Dev.
So there's new collision functions, but did the previous collisions change in any way? Because they're not behaving like they used to. My character has different states depending on whether or not it's colliding with the floor, and that affects both its animations and the actions it can perform. Before the update, it worked fine. With the newest update, the character changes states constantly (from standing to jumping) while running forward, as if it was getting pushed away from the floor for a few frames. A huge chunk of the code relies almost entirely on collision_line, if that's relevant.

EDIT: Something stranger is going on here. I installed 2.1.4 once again and the problem persists. I opened an older version of the project in 2.1.4, and the problem persists. I never had this issue before, but now it's as if the code never worked in the first place.

EDIT 2: Never mind that, I was still using the 2.1.5 runtime, so yes, this new update simply broke collisions in my project.
Can you please file a report with your project: https://accounts.yoyogames.com/contact-us#ingame Thanks
 

jonjons

Member
got an weird error here:
" RetrieveJSON - failed to find: C:\Games\GM2_Proj\inventory\inventory\sounds\snd_menu_select1\snd_menu_select1.yy
finished."


i dont have the file " snd_menu_select1.yy " gm might be refering to the original name, ive renamed all my sound files to 3 letters after i install them ex: " snd_ms1.yy "
 

stigmund77

Member
The compiler seems to be broken for me, once it gets to (Writing Chunk... SOND) the compiler completely breaks and i am unable to use this version of game maker.
 

Dan

GameMaker Staff
GameMaker Dev.
I have this exact same problem... It doesn't happen with all my projects. The compiler gets to "Writing chunk... SOND." And then I get the super non-descript crash you mention above. Going back to an older version of the project does not solve the problem which means it's definitely caused by the update.

EDIT: Reverting to the previous run-time fixes it for now, but sadly bars me from the new functions.
Please send us one of your projects which does this in a bug report - https://accounts.yoyogames.com/contact-us?#studio2 - or, even better, isolate some of the sound files which cause this and send us a project with just them in. Thanks.

EDIT: Can see a new post has appeared above this one whilst I was writing mine, which is the same issue. Can you send us a sample also, please?

EDIT 2: A fix has been found. https://forum.yoyogames.com/index.php?threads/gms2-version-2-1-5.50985/page-2#post-312746
 
Last edited:

sman

Member
And running on iOS I get "linker command failed with exit code 1"
OK, once again, I can't work :( (I believe I spend 90% of my time trying to make the tools I use working :eek:)
 
Y

Yazuka

Guest
And running on iOS I get "linker command failed with exit code 1"
OK, once again, I can't work :( (I believe I spend 90% of my time trying to make the tools I use working :eek:)
Can you please put in a support ticket with your full compiler log, the version of Xcode you are using, the version of iOS your device is and whether you're using VM or YYC.

Thanks in advance.
 

rmanthorp

GameMaker Staff
Admin
GameMaker Dev.
I'm getting similar results, rectangle and rectangle+rotation gives the same results when I rotate an object.

However, for me, the results seem to vary based on the collision function used.

If I use place_meeting() function, the collision is based on the non-AABB collision box.

If I use collision_point_list() function on the same object, the collision is based on the AABB rectangle, the same as you would get if you used the bbox_ variables.

I checked the docs for the two functions, they both say that if you are not using precise collisions, the functions will default to the bounding box.

However, as I mentioned above, place_meeting() uses the new non-AABB rectangle if the instance is rotated, where collision_point_list seems to use the old bounding box AABB method.

Not a deal-breaker for me, just interested to know if this is intended or not.
This is not correct: https://bugs.yoyogames.com/view.php?id=29858 issue has been addressed for the next update. Thanks.
 
C

Cowlord

Guest
Getting Syntax error:

LibGit2Sharp Exception: Object not found - no match for id (45ae9c6ef4696a83db06eac4fba49482824e79ee)
Game seems to still function as it should.
 

Dan

GameMaker Staff
GameMaker Dev.
For any of you with the audio issue in the asset compiler, a sample was filed and we've worked out the issue - for some reason, your sounds are not set to any audio group (not even the default one), and the asset compiler isn't catching this and showing you a message asking you to fix the sound. If you check your sounds and fix any which have the audio group value currently blank, this will fix your project and you should be able to build fine.

https://bugs.yoyogames.com/view.php?id=29859
 

jonjons

Member
For any of you with the audio issue in the asset compiler, a sample was filed and we've worked out the issue - for some reason, your sounds are not set to any audio group (not even the default one), and the asset compiler isn't catching this and showing you a message asking you to fix the sound. If you check your sounds and fix any which have the audio group value currently blank, this will fix your project and you should be able to build fine.

https://bugs.yoyogames.com/view.php?id=29859
great, this worked for me, had the snd_ms1 on a blank audio group
 

gnysek

Member
I'm getting similar results, rectangle and rectangle+rotation gives the same results when I rotate an object.

However, for me, the results seem to vary based on the collision function used.

If I use place_meeting() function, the collision is based on the non-AABB collision box.

If I use collision_point_list() function on the same object, the collision is based on the AABB rectangle, the same as you would get if you used the bbox_ variables.

I checked the docs for the two functions, they both say that if you are not using precise collisions, the functions will default to the bounding box.

However, as I mentioned above, place_meeting() uses the new non-AABB rectangle if the instance is rotated, where collision_point_list seems to use the old bounding box AABB method.

Not a deal-breaker for me, just interested to know if this is intended or not.

OK, So I've got reply from helpdesk. It seems, that the collision mask of colliding instance decides what will be used. When I used ellipse, it enables per-pixel collision on both, when I used "rectangle" on my circle sprite, it seems to work as described:upload_2018-8-3_12-40-22.png


So, "rectangle" and "rectangle + rotation" works as long, as another object also uses one of those collision types.
 

GMWolf

aka fel666
OK, So I've got reply from helpdesk. It seems, that the collision mask of colliding instance decides what will be used. When I used ellipse, it enables per-pixel collision on both, when I used "rectangle" on my circle sprite, it seems to work as described:View attachment 19937


So, "rectangle" and "rectangle + rotation" works as long, as another object also uses one of those collision types.
That seems... Rather random.
Good to know though!
 

rIKmAN

Member
When I click the Run button for Android, nothing happens, not even in the Output window... :(
Reverting to the previous run-time changes nothing.
Sadly this something to do with your system or setup specifically, rather than the new update causing the problem - as evidenced by the older runtime not working either.

I tested Android yesterday and it worked fine, and have also helped someone else in another thread on here to compile successfully with API28.

This is not correct: https://bugs.yoyogames.com/view.php?id=29858 issue has been addressed for the next update. Thanks.
Things like this slipping through are why QA have a poor reputation.
It's hard to understand how a new feature wasn't tested enough that this basic issue wasn't caught, it's not like this is a hard to find edge case.

Working together with the community to allow selected members to join some sort of private beta testing group and give feedback to QA to confirm and act upon would benefit everyone in the long run.

Something you should at least consider looking into, I think.
 

sman

Member
Sadly this something to do with your system or setup specifically, rather than the new update causing the problem - as evidenced by the older runtime not working either.
I've tested Android with the previous IDE and the previous runtime and it worked.
Also, when I use the new IDE, the fact that nothing happens when I click the run button is a problem : the IDE should have given me a message to explain the problem.
More: with the new IDE, when I press F5, it crashes!
 
M

MarceloP

Guest
I'm getting similar results, rectangle and rectangle+rotation gives the same results when I rotate an object.

However, for me, the results seem to vary based on the collision function used.

If I use place_meeting() function, the collision is based on the non-AABB collision box.

If I use collision_point_list() function on the same object, the collision is based on the AABB rectangle, the same as you would get if you used the bbox_ variables.

I checked the docs for the two functions, they both say that if you are not using precise collisions, the functions will default to the bounding box.

However, as I mentioned above, place_meeting() uses the new non-AABB rectangle if the instance is rotated, where collision_point_list seems to use the old bounding box AABB method.

Not a deal-breaker for me, just interested to know if this is intended or not.
OK, So I've got reply from helpdesk. It seems, that the collision mask of colliding instance decides what will be used. When I used ellipse, it enables per-pixel collision on both, when I used "rectangle" on my circle sprite, it seems to work as described:View attachment 19937

So, "rectangle" and "rectangle + rotation" works as long, as another object also uses one of those collision types.
That's ultra important to know! I really think that it should work with all "non-precise" collision types, since the documentation states that precise and non-precise are the only 2 different types of collisions we should care about when checking for collisions... Anyway, I hope they fix it soon.

Thanks!
 

Ricardo

Member
I want to put this "fix" into discussion. game_display_name was always lowercase in HTML5. Now it was "fixed" and is case sensitive, respecting what you have in Options>Browser Title. This seems like a small change, BUT, considering game_display_name is used in the Local Storage files keys, this change has a big potential to break saves in HTM5L games.
 

rIKmAN

Member
I've tested Android with the previous IDE and the previous runtime and it worked.
Also, when I use the new IDE, the fact that nothing happens when I click the run button is a problem : the IDE should have given me a message to explain the problem.
More: with the new IDE, when I press F5, it crashes!
You said "Reverting to the previous run-time changes nothing." which i took to mean the previous version didn't work either.

I'm not saying you don't have an issue, I'm saying that I tested Android last night and everything worked fine, as did someone else who I helped use API28 to compile so the issue isn't a widespread problem with the IDE in general.
If it was then we would all have the same complaint that you have.

It seems like it may be specific to your system somehow, whether that be drivers, SDK versions, cached files etc.
Try fully uninstalling GMS2 from your system, rebooting and installing the new version again and see if that helps.
 
F

Fiskyfisko

Guest
Oh my god my game is running on 1500fps now instead of 1100, what did you dooo.... The collisions are behaving differently again but it's not that bad.
 

Maa2007

Member
hello so after i updated my game maker to the latest version which is 2.5, the game which used to work on my android tablet is not working any more, after building it, the game still installs and play on my phone, but only get installed on my tablet, and when trying to open the game it give me this massage:

unable to find library for this devices architecture, which is armv71, ensure you have included the correct architecture in your APK


and yes i have the armv71 check-boxed in android options
 

Attachments

Mick

Member
The list of installed runtimes seems to be missing after updating to version 2.1.5, or am I missing something (if you pardon the pun)?

upload_2018-8-3_22-39-48.png
 
Getting an error message on running any game. Machine starts to chug, screen goes white and I get the following error:

Win32 function failed: HResult: 0x887a0005
Call: GR_D3D_Device->CreateBlendState as line 467 in file \StateManagerM.cpp

Cannot compile anything. Ideas?
 
Does this happens with a fresh empty room project?
It doesn't error on using a 'clean' project.

All previous projects now fail like this.

If I open a project (one from a previous version) and get the error then switch to the 'clean' project it has the same error.
 
M

Monky

Guest
Hey, quick question. I wanna report a bug I have on 2.1.5, is it okay if I file it under 2.1.4? It's the only option in the Select IDE/Runtime sections on the form.
 
E

Excalipur

Guest
For any of you with the audio issue in the asset compiler, a sample was filed and we've worked out the issue - for some reason, your sounds are not set to any audio group (not even the default one), and the asset compiler isn't catching this and showing you a message asking you to fix the sound. If you check your sounds and fix any which have the audio group value currently blank, this will fix your project and you should be able to build fine.

https://bugs.yoyogames.com/view.php?id=29859
A hundred thanks! This solved my problem. Everything seems to be working now.
 

Neverdusk

Member
So I've been loving the updates, but the new double-click requirement to open object events is really hurting my workflow. I'm used to bringing up an object and rapidly opening the events I need to immediately access them, but since the update double-clicking has felt unnecessarily slow and awkward. Will there be any chance of a toggle for single-clicking and double-clicking to open object events? I understand that the change was made to improve workflows and prevent copy-paste mistakes according to the release notes, but this little change has felt like a major backstep for me. That said, I'm really grateful for everything else, and the new collision scripts are very exciting!
 
P

psyke

Guest
So I've been loving the updates, but the new double-click requirement to open object events is really hurting my workflow. I'm used to bringing up an object and rapidly opening the events I need to immediately access them, but since the update double-clicking has felt unnecessarily slow and awkward. Will there be any chance of a toggle for single-clicking and double-clicking to open object events? I understand that the change was made to improve workflows and prevent copy-paste mistakes according to the release notes, but this little change has felt like a major backstep for me. That said, I'm really grateful for everything else, and the new collision scripts are very exciting!
Same workflow problem here, they should at least add an option in the preferences for us to choose to open the events with one click, two clicks is like twice the effort in some situations, specially when you need to get some code from other events or to do quick updates.
 
G

Guest User

Guest
The list of installed runtimes seems to be missing after updating to version 2.1.5, or am I missing something (if you pardon the pun)?
windows? mine are showing up there as expected, so i think you are experiencing an issue.
 
L

Ludorverr

Guest
Same workflow problem here, they should at least add an option in the preferences for us to choose to open the events with one click, two clicks is like twice the effort in some situations, specially when you need to get some code from other events or to do quick updates.
I think triple-click would be superior, after all we don't want people accidentally double-clicking things. It's only 1 extra click, so no one should be against triple-click because it's not a big deal. :)
It's not just events, it's the entire resource tab that is absolutely sluggish. FL Studio, Photoshop and every single other program's user interface all operate in the same manner, the single-click manner. The only example of double-clicking I can even think of is executing programs on the windows desktop. Or how about internet browsers in general, imagine double-clicking links just to browse around the internet. We do this all day long every day, it's the little things that add up.

Something interesting I've noticed about the resource tab is that you can hold middle-click (or right-click) to move resources around without opening them. Hmm. It's almost as if there's no reason whatsoever to need to single-click highlight resources.

From now on every time you use a user interface in ANY program or game I want you to count how many times you clicked to perform the action you wanted. Take note of all the people complaining about accidentally clicking on those things too. This "accidental clicking" is an imaginary boogieman invented only by the game maker community.
 
D

Dawn

Guest
Unless they fix or add an option for the double click bullshittery which I don't really expect to happen anytime soon at all, I'll have to reinstall 2.1.4.. Or I may have to make an autohotkey macro dedicated to opening object events. Again, nice work.
 

MissingNo.

Member
So the update seems to have broken my scipt-based state system. In my state system I have a dodge-roll state which used to work whether the player was moving or not, but now it doesn't work if the player is idle/not moving, and won't work at all if the sprite assigned to sprite_index only has one image frame. After doing some experimenting, my guess is the image_index function is bugged in some way?

EDIT: It seems animation definitely isn't working correctly if the sprite only has one image frame. Could possibly be an issue with image_index, animation itself, or the animation end event.

EDIT 2: I've reported this as a bug, hopefully it gets fixed soon.
I'm definitely having something similar happen I made a thread with more details:

https://forum.yoyogames.com/index.php?threads/latest-runtime-causing-a-bug.51067/

Basically it seems animations with only one frame cause "animation end" to immediately fire as soon as the sprite is applied.

I reverted back to runtime 2.1.4.218 to fix the issue.

I guess I won't need to file a bug report if one is already been made.
 

Toque

Member
I think triple-click would be superior, after all we don't want people accidentally double-clicking things. It's only 1 extra click, so no one should be against triple-click because it's not a big deal. :)
It's not just events, it's the entire resource tab that is absolutely sluggish. FL Studio, Photoshop and every single other program's user interface all operate in the same manner, the single-click manner. The only example of double-clicking I can even think of is executing programs on the windows desktop. Or how about internet browsers in general, imagine double-clicking links just to browse around the internet. We do this all day long every day, it's the little things that add up.

Something interesting I've noticed about the resource tab is that you can hold middle-click (or right-click) to move resources around without opening them. Hmm. It's almost as if there's no reason whatsoever to need to single-click highlight resources.

From now on every time you use a user interface in ANY program or game I want you to count how many times you clicked to perform the action you wanted. Take note of all the people complaining about accidentally clicking on those things too. This "accidental clicking" is an imaginary boogieman invented only by the game maker community.
Never seen a complaint about single click.
Seriously we are adding clicks to improve workflow???
We should be improving these type of things. Removing clicks and redundancy not adding.
What?? I have no other software in my workflow that does that.
Was it double click in the past? People wanted this??

Please reconsider!!!
 
Last edited:

Smiechu

Member
Never seen a complaint about single click.
Seriously we are adding clicks to improve workflow???
We should be improving these type of things. Removing clicks and redundancy not adding.
What?? I have no other software in my workflow that does that.
Was it double click in the past? People wanted this??

Please reconsider!!!
The problem is not the double click, the problem is lack of a setting enabling/disabling it.
 

JeffJ

Member
This is, once again, a good example of "if you want these bugfixes and/or this new feature, you also have to take this 💩💩💩💩ty thing that we made that has nothing to do with what you want".

What I really don't get is why it wasn't just made into a preference and then just randomly thrown in there. So now it's either accept it or don't update.
 

hth

Member
Good morning.

So my project is a bit 💩💩💩💩ed after the update to 2.1.5.

I created a thread two days ago in which I explain in a bit more detail what goes wrong, what I tried to narrow the issue down and possibly find a solution. (Spoiler: Didn't do the latter.) I'll give a quick tl/dr here:
  • In my main project, which has reached a certain level of complexity with around 10.000 lines of code and ran/runs perfectly fine on 2.1.4, suddenly EVERY call to ANY collision function (place_meeting(), position_free(), line_collision(), what have you) will make Runner.exe exit ungracefully. No further info to be had on what goes wrong where.
  • Doesn't matter what object is checked for, doesn't matter what its sprite's collision mask is.
  • Removing all objects that do collision checks from the project will stop the game from crashing, running perfectly stable doing AI calculations, physics and everything else except collision checks. Then inserting one simple
    Code:
    var foo = place_meeting(x, y, obj_wall_parent);
    in ANY object, like the physics manager or the camera object, will have the game reproducably crash again.
  • Collision checks seem to be fine in events that run only once, like Create or Room Start, but make the game explode on first (!) function call in repeated events such as the Step and Draw ones. To make it weirder, there is a tiny chance that the same code WILL run. Like, hitting F5/F6 a hundred times, one time it'll just work. As this happens so rarely, I didn't have the patience to investigate further, but I suspect the game would crash then as well if any more collision checks were made. On the rare occasions that it did happen, there was only ever one call followed by some debugging output and game_end().
  • Fresh projects with basic collision checks work fine, calling all kinds of collision functions in a ten-minute while loop also worked fine in a fresh project. Existing smaller projects like sandboxes for experimenting with game concepts also run as expected.
  • I should also add that, yes, my system is stable.
After two precious days wasted, I'll be rolling back to 2.1.4 and continue development. Naturally, I still hope to some day find out what the problem might be.

Edit 1: What I haven't done yet: Run the code through the YYC compiler. Should probably do that and report back.
Edit 2: Did that now. Two minor (and unrelated) compiler warnings, fixed those, same result.
 
Last edited:

Toque

Member
This is, once again, a good example of "if you want these bugfixes and/or this new feature, you also have to take this ****ty thing that we made that has nothing to do with what you want".

What I really don't get is why it wasn't just made into a preference and then just randomly thrown in there. So now it's either accept it or don't update.
What’s the rational to even add clicks there?

They are so busy doing serious bugs and demanded features that it’s unlikely they will change it.

I appreciate the effort on other fronts.
Nice to see updates.
Thank you.
 

IICIIEII

Member
Hello!

I hope this is the right place to post this :). I am experiencing a problem with the new update and I filled a ticket about it, but in the meantime I am very curious to see if somebody else has the same issue. In my case, I have an enemy AI chasing system based on mp_grids. This is the creation code:

//Room grid initialisation
var cell_width = 3; //grid horizontal pixel size
var cell_height = 3; //grid vertical pixel size

var hcells = room_width div cell_width; //number of horizontal cells, in integer number
var vcells = room_height div cell_height; //number of vertical cells, in integer number

global.grid = mp_grid_create(0, 0, hcells, vcells, cell_width, cell_height); //create the grid

//Objects to avoid in the maps
mp_grid_add_instances (global.grid, obj_collision, true);
mp_grid_add_instances (global.grid, obj_enemy_collision, true);


and then, when the enemies are actively chasing you, they call this bit of code (Kara is the main character in the game and her coordinates are obj_kara.x, obj_kara.y)

//move the enemy through the path chasing Kara
if (mp_grid_path(global.grid, enemy_path, x, y, obj_kara.x, obj_kara.y, 1))
{
path_start(enemy_path, enemy_speed, path_action_stop, false);
}


the script is more complex (it changes sprites in real time and more things) but that is the important bit.

Until the last update, I was able to have 23-25 enemies at once chasing my main character in a 3000x3000 room at, at least, 300 fps. However, with the new version frame rate dropped to 9fps, so I had to reduce the number of enemies to three and still, I am getting around 70 frames per second. Do you have any idea why this could be happening? My enemies are not really checking for collisions in the map and are not solid, although they have a square collision mask. They just avoid the forbidden objects in the grid.

My computer is an I7 2600K with 32Gb and a GTX980.

Thanks!! :)
 
What I really don't get is why it wasn't just made into a preference and then just randomly thrown in there.
Probably (as I mentioned in another post) because the fix is to require double-click, and if they enabled people to single click again through a preference it would lead to the potential of more people logging the original bug and complaining about it in exactly the same way people are now complaining about the fix.

Seems to be a no-win situation for YYG here: either have the bug and have people complain about it, or fix the bug and have people complain about the fix.

Personally I don't mind whether it is single-click or double-click. Only thing I found with the single click was that I might accidentally click on the wrong event and I wouldn't want the code window to appear, so now that doesn't happen as I have to double-click.
 

IICIIEII

Member
Hello again!

I am extremely sorry, everything is solved now and it had nothing to do with the new Game Maker version, but me accidentally erasing a portion of the code:

In the enemy create event I was supposed to have:


randomize()
path_interval = irandom_range(10, 40); //when is the path recalculated in fps
path_frame_counter = 0; //variable to add path interval on each frame
enemy_path = path_add();


so when you call the chasing script you don´t need to run it every frame for all the enemies:

//move the enemy through the path
path_frame_counter = path_frame_counter + 1;


if (path_frame_counter >= path_interval) //if we are above the path interval
{
if (mp_grid_path(global.grid, enemy_path, x, y, obj_kara.x, obj_kara.y, 1))
{
path_start(enemy_path, enemy_speed, path_action_stop, false);
}
path_frame_counter = 0;
}


However, I accidentally erased path_frame_counter and removed the condition for some tests, not restoring the code after that. This means that every single enemy was calculating its path once per frame, and therefore the framerate was so poor.

So sorry again!
 
G

Guest

Guest
An Android YYC build now takes a bit over 10 minutes; before, it was 4 minutes, which felt long. But I don't immediately see anything broken, so that's good.
 
The list of installed runtimes seems to be missing after updating to version 2.1.5, or am I missing something (if you pardon the pun)?

View attachment 19942
You're using the VS Studio for GML, nice. However, I was having the same problem and it turned out to be a problem with the VS skin. You're going to have to go back to the original GM skin and check the runtimes from there.
 
Top