Major Updates to Mod Scripts Coming in March

That's completely fine, I don't mind talking about programming on any level, even the ground floor.  The people who don't struggle with imposter syndrome from time to time don't know enough about the subject to know how much they don't know yet. =D  

Hey everyone! It's a few weeks later than we intended, but tomorrow is the big day! Wednesday, April 17th at 1:00pm PDT, the updated Roll20 Mod Server will be rolled out! We aren't expecting any problems, but to be prepared for good experiences in peak times, we're starting up early enough that we can monitor games as things get moving. Your feedback will be important to make sure we identify any issues early and get a good resolution to them quickly.  If there ends up being a problem, we can safely revert back to the prior version with no loss of data or configuration, so keep us posted about your experiences. I'm looking forward to seeing what we all build together, see you tomorrow!

Hey everyone! We're going to pull this back.  We gathered some great data, but there are some memory constraints with very large games that we need to solve.  While we would be able to support 99% of current games, there would be some edge cases where games would not be able to start and that's just not tenable.  I'm going to be working on optimizing those memory constraints, and see about speeding up startup as well.  We'll post back with a new schedule when we get a handle on the changes.

Excellent! I’m keen to help with Beta testing. I missed the rollout because I was asleep in Australia at the time. 

Okay, I wanted to update the old Teleport mod, to do so, I believe I need checkin-checkout software to work with the mod server, but I'm short on details and it's been a few years since I've done it. Is there a quickstart around, and has it changed? 

What tag do you want us to use in the help request?

When you start testing by copying a game without players , then adding Mod scripts without first launching the game, expect errors on mods that access the player objects. For example: adding APIHeartBeat yields TypeError: Reduce of empty array with no initial value. Launching the game once solves the issue .

Oh, interesting!  I'll try and duplicate that. 

Currently you can identify that the Experimental sandbox is selected because the Mod Output Console shows "SANDBOX: Ready fired after 6.24s, 578 objects." would it be possible to also report whether jumpgate beta is applicable for this game? I am comparing four versions (legacy vtt, jumpgate, legacy mod, new mod) and having this in the Mod Output Console would make it easier comparing logs.

P1) &nbsp; The new findObjs implementation does not behave as the old one in exactly&nbsp; the same way. For example: searching on imgsrc must have had some extra logic to compare image urls. this might be a reason for the many macro creations of ApiHeartbeat. The CombatMaster mod also creates many objects too many. 2) The new toFront does not bring all objects to the front, it must have some logic to test the controlledby field. 3) iobserved that the new sandbox code checks image urls and changes these to the new image&nbsp; delivery urls. <a href="https://s3.amazonaws.com/files.d20.io" rel="nofollow">https://s3.amazonaws.com/files.d20.io</a> /... =&gt;&nbsp;&nbsp; <a href="https://files.d20.io" rel="nofollow">https://files.d20.io</a> /... Any mods that compare img url fields (with or without getCleanImgsrc) break. Details and investigation details&nbsp; have been pm-ed to the Aaron. BTW i am happily surprised that the bugs i run into are specific edge cases. This is a good first public beta.

I can report that all the Mods I use in my 13th Age Glorantha game are working exactly as expected. They are: MetaScriptToolbox TokenMod TokenNameNumber ScriptCards I noted Martijn's report above and launched once before adding Mods and switching to Experimental. Next I'm going to copy my 13G &amp; TOR2e Library games and switch to Experimental for more testing.

I just tried using a ScriptCards feature that I haven't tried before, and I got an error, so I'm reporting it here and in the ScriptCards thread.&nbsp; The script I'm running is&nbsp; !scriptcard {{ &nbsp; &nbsp; --#title|Style_Runes &nbsp; &nbsp; --/| Rune-Air (Orange) &nbsp; &nbsp; --#titlecardbackground|#FFA500 &nbsp; &nbsp; --S#Rune|Air }} and the error message I get is&nbsp; "SANDBOX: Ready fired after 4.39s, 1138 objects." "WARNING: Refusing to set \"current\" to undefined. [Roll20 attribute -NwZEHmFt-tTY2RhdogE]" I was using the&nbsp; Original Method of Saving and Loading but I wanted to try the&nbsp; Enhanced Methods of Saving and Loading as part of my testing.&nbsp;

Martijn S. said: Currently you can identify that the Experimental sandbox is selected because the Mod Output Console shows "SANDBOX: Ready fired after 6.24s, 578 objects." would it be possible to also report whether jumpgate beta is applicable for this game? I am comparing four versions (legacy vtt, jumpgate, legacy mod, new mod) and having this in the Mod Output Console would make it easier comparing logs. There isn't a way for the Mod API server to tell if Jumpgate is enabled that I'm aware of.&nbsp; The Mod API server operates on the data model of the game, where as Jumpgate is a different UI that uses the same data model.

Martijn S. &nbsp;said: P1)&nbsp; The new findObjs implementation does not behave as the old one in exactly the same way. For example: searching on imgsrc must have had some extra logic to compare image urls. this might be a reason for the many macro creations of ApiHeartbeat. The CombatMaster mod also creates many objects too many. That's right!&nbsp; When I cleaned up the image code, and added in support for duplicating marketplace content already in the game, I set up optimization of the urls to get rid of the problem of aways needing to substitute a thumb url, and changed them to the faster image endpoint.&nbsp; However, as you surmised, that's causing a miss on finding the image.&nbsp; Internally, I'm matching the imgsrc in other places against a uniq key calculated out of the URL.&nbsp; I don't know if I can repurpose that code, but at a minimum, I can do the same transform on the incoming url before the match and that should handle the problem.&nbsp; 2) The new toFront does not bring all objects to the front, it must have some logic to test the controlledby field. I will dig into this one.&nbsp; Are you saying it only seems to work on objects you control?&nbsp; Are you using TokenMod for this by chance? 3) I observed that the new sandbox code checks image urls and changes these to the new image delivery urls. <a href="https://s3.amazonaws.com/files.d20.io/" rel="nofollow">https://s3.amazonaws.com/files.d20.io/</a>... =&gt;&nbsp; <a href="https://files.d20.io/" rel="nofollow">https://files.d20.io/</a>... Any mods that compare img url fields (with or without getCleanImgsrc) break. This is the same issue as 1), really.&nbsp; I'll get that addressed. Details and investigation details have been pm-ed to the Aaron. BTW i am happily surprised that the bugs i run into are specific edge cases. This is a good first public beta. Thanks on both counts!&nbsp; It would be great to release with no bugs at all, but as they say, "if debugging is the process of taking bugs out of code, coding is the process of putting them in there!"

Andrew R. said: I just tried using a ScriptCards feature that I haven't tried before, and I got an error, so I'm reporting it here and in the ScriptCards thread.&nbsp; The script I'm running is&nbsp; !scriptcard {{ &nbsp; &nbsp; --#title|Style_Runes &nbsp; &nbsp; --/| Rune-Air (Orange) &nbsp; &nbsp; --#titlecardbackground|#FFA500 &nbsp; &nbsp; --S#Rune|Air }} and the error message I get is&nbsp; "SANDBOX: Ready fired after 4.39s, 1138 objects." "WARNING: Refusing to set \"current\" to undefined. [Roll20 attribute -NwZEHmFt-tTY2RhdogE]" I was using the&nbsp; Original Method of Saving and Loading but I wanted to try the&nbsp; Enhanced Methods of Saving and Loading as part of my testing.&nbsp; This is great, actually.&nbsp; There is a problem in the existing sandbox when attribute values get set to a bad value (particularly NaN).&nbsp; What this means is that ScriptCards has a case where it tries to set the value of current to undefined, and the new Mod Server is detecting that and not doing it.&nbsp; Right now, that means that current goes unset, and retains its prior value.&nbsp; It might make more sense in these cases to just set it to "" (empty string), assuming that a bad value means just clear it.&nbsp; I didn't make that choice initially because to my thinking it would always be a bug in the script itself if this came in as a bad value, and I'd want to find and fix that on the script side.&nbsp; Making it vocal with a warning like this should let script authors clean up those problems without breaking the sandbox when it happens.

@The Aaron, one of the challenges I faced with CM is being able to transfer the existing combat to a new page.&nbsp; The current exposed architecture of Roll20 doesn't allow for changing the current page.&nbsp; I looked at the pagefolder functionality.&nbsp; Could this be leveraged?&nbsp;

PageFolders won't help with that, but it is something I've talked to Riley about.&nbsp;

Thank you for the explanation, and I agree with your reasoning. I found the problem with my script and corrected it.&nbsp; The Aaron said: This is great, actually. … Making it vocal with a warning like this should let script authors clean up those problems without breaking the sandbox when it happens.

Great!

This just happen. has anyone encounter the same thing.

Just in case somebody ask this is the full view. I have scripts on this campaign.

Never mind it corrected by itself.&nbsp;

I keep getting errors in one of my campaigns where the sandbox frequently crashes with not a informative error message. could someone help me out? Thanks.

Is there an error log below (see an example in the screen shot two posts above yours)? That might contain useful information to include here.

Possible Infinite Loop Detected happens when the scripts take too long before relinquishing control back to the sandbox.&nbsp; Usually, this will be a script reacting to an event by doing something that takes a very long time.&nbsp; On rare occasions, this will happen when you start the sandbox if your game is extremely large (like, more than 500,000 objects).&nbsp; So, the two important questions are: Is your game really large? What scripts do you have installed? In the case of 1, you can move things out of the game to another game for storage and remove them from the source game.&nbsp; In the case of 2, you need to identify which script might be causing the issue and get it fixed.&nbsp; The fix is often pretty easy, and many of us can help with that.&nbsp; Just post a list of the scripts you have, and we can help you start looking into it.

I dont think its to large but I could be wrong. I can invite to game if necessary.

Aryan said: I dont think its to large but I could be wrong. I can invite to game if necessary. with the survey mod you can count the objects in your game and know for sure. See <a href="https://app.roll20.net/forum/permalink/10675897/" rel="nofollow">https://app.roll20.net/forum/permalink/10675897/</a> Let's go through your list of MODs. Most of these only react to chat or react to changes to single objects (and then change single objects). The way to detect whether these MODs are the culprit is to have a second browser open with the MOD Output Log next to your play game. Aura/Tint HealtColors creates some FX object on startup. As FXs are documented to be not implemented, that mod might be suspect. The '5th Edition OGL by Roll20 Companion Script' MOD sets the on('chat:message',..) outside the on('ready',...), that might give an issue if you chat something at exactly the moment that the sandbox is starting up. But as you state it is recurring, this is very unlikely.

with that list of scripts, there isn't an obvious culprit.

Nice updates!!