Warzone 2100 Forums

Zepherian has discovered a critical bug: Vtols refuse to reload on pads in the latest buildbot version. I could reproduce the problem.

Forgon wrote:Zepherian has discovered a critical bug: Vtols refuse to reload on pads in the latest buildbot version. I could reproduce the problem.

Can confirm. They try to guard the VTOL pad instead.

Berserk Cyborg wrote:

Per wrote: I have some notes of my own for the changelog, but I think you should commit this first, then I can modify yours as appropriate.

Ok, I pushed it in 7e9a52df8a4bd4baa7fc12252c89e32c16e3291f.

I'm still of the opinion that we should call the new version 3.3. There are many important changes and most of all you can't play 3.2.3 saves with the latest master due to all these changes. I think this is too much for a small step from 3.2.3 to 3.2.4.

Berserk Cyborg wrote:

Forgon wrote:Zepherian has discovered a critical bug: Vtols refuse to reload on pads in the latest buildbot version. I could reproduce the problem.

Can confirm. They try to guard the VTOL pad instead.

I don't think it tries to guard it. It just uses a poor mouse icon. I broke this when removing the cluster code. Fix in e3ea0d9eee64259009ba02b797e76acf4781c900 (hopefully). The rearm code is really awful

NoQ wrote:...but the rest is completely personal-taste-driven.

But can any of you do this with Python? And yes, it does work.

And ever wonder how bugs are fixed for wz2100? Joking aside, I think this next release will be amazing.

Berserk Cyborg wrote:

NoQ wrote:...but the rest is completely personal-taste-driven.

But can any of you do this with Python? And yes, it does work.

Doesn't python code normally get pre-compiled into .pyc bytecode before getting loaded, yielding much better results than minification? (That said, i'm not convinced that python is "faster" than any of the modern javascript engines. But i was mostly talking about splitting up into files when i said "taste-driven".) Actually, for our JS scripts it is likely to be much better to "JIT" them completely into binary code before loading, because we don't care much about loading time (loading game models is much slower anyway), but we do care a lot about lags and stuttering during the game. See also WebAssembly.

I was thinking that we should move the campaign testing thread to the new development/campaign sub-forum. That way an official release topic does not get buried and it makes more sense to have it in a development forum than being in news/announcement.

Feel free to move the thread.

Also, I started adding some debugging tips and hints in e4a1139918687a8da912e7a06d8c677731e07854. I realized I have little idea of how to debug campaign anymore. So please fill in more information and improve things.

Per wrote:I realized I have little idea of how to debug campaign anymore.

We've got http://developer.wz2100.net/wiki/jscam# ... gingtricks and also libcampaign dumps a lot of stuff into console and highlights scripted areas automatically when debug mode is on.

There is also camTrace which outputs stuff to the terminal when in debug mode. I will try updating that document soon.

Why assert in loadLabels() with dead objects in a group label? I think it would be fine to just ignore any dead objects in those labels. Some mission label files use group labels and if a player destroys part of the group they will be welcomed with a noisy assert on load.

Surely dead objects should never be saved to savegames in the first place?

Per wrote:Surely dead objects should never be saved to savegames in the first place?

They aren't, at least when I searched a save file for object IDs. It's just in the past I encountered groups having a count less than zero when used with the "group" map label.

Ah. Hmm. There's something about the whole label group and script group saving/loading which doesn't quite look well designed in how they interact, in the sense that if you remove droids from label-created groups during the game, they will be re-added on load... And dead droids will be saved to label files without hesitation.

Anyways, feel free to remove the assert for now.

Zepherian has discovered that VTOLs

• still fail to rearm: Few or none of them will land on pads even if explicitly ordered to do so.
  After returning from an attack, many fly to the HQ.
• can fly into buildings at high altitudes and get stuck there.

Did anybody pay attention to this is not the correct mapping?

Also there is the problem of shadows. Vincent said that he fixed it.