Huge Jungle Project

Monkey Forums/Monkey Programming/Huge Jungle Project

ziggy(Posted 2014) [#1]
If anyone has any very big Jungle Ide project (several hundreds of files if possible) and want to help Jungle Ide development, I need to profile a new code analyzing tool on Jungle Ide before next release, but I would need a big project (ore more) to ensure it works as it should.
This new analyzer will provide useful information such as local variables shadowing, overloading usage and warnings when an overload is not case correct, etc. This is done using several threads but I need to be sure it's not slowing down the IDE when used on very big projects.

Thanks!


CopperCircle(Posted 2014) [#2]
I have some very large client projects, I can't provide the actual projects but could run some tests for you.


ziggy(Posted 2014) [#3]
Ok, I'll send you a beta in some days if it's of for you. Thanks


muddy_shoes(Posted 2014) [#4]
I don't understand the question really. It's trivial to generate a Jungle project of hundreds of files just by using a few modules like box2d and diddy and including the modules in the project. Is that what you mean? Also, to be frank, I really don't care about the feature being discussed in comparison to fixing existing issues like auto-completion's failure to handle generics.


ziggy(Posted 2014) [#5]
I don't understand the question really. It's trivial to generate a Jungle project of hundreds of files just by using a few modules like box2d and diddy and including the modules in the project. Is that what you mean?
I already did that, but I want to test this in code written by more people. Sometimes a bug may show when you analyze a specific code structure you've not tested before unless you test the thing on several different coding styles during beta testing.

My idea behind this feature is -in a bit longer therm- to be able to add some pragma-like features to the IDE, such as support for overrides or overridable keywords-like in the code, to make the code easier to maintain and to make it easier to find some coding bugs. Typicial errors like hidding an overriden method because Monkey does force to override all methods with the same name to maintain the inheritance chain when extending classes, or using wrong case on a method overload, etc. This analyzer should also detect unused local variables and mark them as such.

Also, to be frank, I really don't care about the feature being discussed in comparison to fixing existing issues like auto-completion's failure to handle generics.
Fortunately, those are not exclusive things and I'm working on improving generics too. Improving generics support has been a priority for some months, but I haven't yet got a fast enough implementation that do not have side effects. Now I'm working on a new approach that seems to be working much better than previous ones, by internaly using generic classes as if they were c++ templates, and generating in-ram copies of the original classes with all generic "resolution" pre-calculated. This seems to be fast enough, but it's not really stable jut yet, as soon as I have it stable enough, and if there are no additional unexpected issues, I'll add this new Monkey parser to Jungle Ide.


muddy_shoes(Posted 2014) [#6]
As a customer (although to be fair one who bought the lifetime license and so is zero future revenue) I'm unhappy with your apparent concentration on creating language-like features over simply supporting the language itself.

There's this feature and the "hiding" feature which suggests you're looking at creating some sort of JungleIDE meta-language. Meanwhile, as an IDE, Jungle's auto-complete falls short, the formatting still breaks regularly, it has no refactoring of any kind, no helper features like add interface/abstract skeletons, no basic navigations like find references or find children and outside tool integration is just starting to get a look in. I'm very sure you make little to no money from Jungle and I'm not demanding anything but I just don't get where you're going.


ziggy(Posted 2014) [#7]
@muddy_shoes: I appreciate a lot your feedback. In short, all the features you miss are reduced to the same issue, which is why I'm trying to improve the generics thing. Providing reliable find usages or refactoring requires the AST to include all generics information, no matter how complex they are, and I'm still trying to get that right. Once I get the internal API to solve all generics details, I would love to add some proper refactoring tools. But it's a slow process as I need to also concentrate in regular support, and other areas of the IDE.

As instance, this Hidden thing, is part of a custom development request by Playniax, for their Ingnition framework and I plan to use it too on JungleGui module too. I understand what you say, and I agree with you, it's not that we are not focussing on that, is that it's not yet ready, we've done several implementations of it, and none was good enough to work reliably and fast. But, obviously, it's important for me and for some Jungle Ide users like you, and I hope to have better support for Generics soon. Then, if generics work as expected, we could get a reliable enough binding solver to provide safe refactoring tools like that ones you're describing, which is something I would love to add to Jungle Ide.


rIKmAN(Posted 2014) [#8]
I've asked this before on the jungle forums, but I would LOVE to be able to lock the target selection so it cannot be changed.

The amount of times I am clicking/scrolling around editing images, sounds, data files or whatever (dual screen too) and go back to Jungle and hit F5 and find I am trying to compile for Ouya or Windows8...well it must be in the thousands now.

There must be a way to allow us to choose a target and then "lock" the selection so no matter what it will not change the target in the selection box.

I think you told me to hide the controls last time, but that doesn't work either as IIRC they still change even though the box was not visible.


muddy_shoes(Posted 2014) [#9]
Most of the time when that happens to me it's because Jungle doesn't remove the focus from the target selection box when you select a target or when you remove focus from Jungle as a whole. So you alt-tab or whatever away from Jungle, come back, try using the scroll wheel and accidentally change the target. It's an annoyance that rates about on a par with tabs being full-screened when I'm just trying to click back to a file because the double click detection is too sensitive and there's no tab scroll function.


ziggy(Posted 2014) [#10]
@rIKmAN: This will be fixed on next stable release.

@muddy_shoes: I've also added an option to disable the double click maximizing, so you can use Ctrl+U instead, or the context menu. That said, I'm using regular windows forms doubleclick events, so not sure why it's not working as any other double click for you.