Silo of Reusable Code (Contributors wanted)

[akamicah]

Hi all,

I'm piecing together a library of reusable code to accelerate application development in Purebasic, as I've tinkered with PureBasic since the very early days and one thing that kinda held me back developing more ambitious projects with it was having to dig around my own mega repository of half baked projects for snippets and bits to incorporate into another project.

With that in mind, I've started this repository on GitHub with a mission to create/procure, refactor and adapt code I stumble across which I find useful and think others will too.

Just like everyone, my time isn't infinite with work and other projects I actively work on, so I'd love it if other members of the community can join me in this mission through posts here or through fork/pull requests.

You can find the repository here:

https://github.com/akamicah/puresilo

Code must be MIT compatible mind to be as friendly to others as possible, and I will refactor/modify/tidy up code if needed to produce a consistent style throughout the library - just something to keep in mind

So far I've added my logging library, also introduced Vector2, Vector3 (of 4 different types) with a bunch of methods, defined Vector4, Rectangle to be expanded too, and will be introducing more as time goes along.

[STARGÅTE]

I just looked over it quickly, but I want to mention some issues:

• You define a function RSqr(Number.d), the reciprocal square root. However, the definition is wrong:

  Code: Select all

    Procedure.d RSqr(Number.d)
      ProcedureReturn 1.0 / Number ; << should be 1.0 / Sqr(Number)
    EndProcedure  

• Similar issue in vector2.pbi, here your normalize procedure misses the square root as well:

  Code: Select all

      Procedure Vec2#Type#_Normalise(*Vec.sVec2#Type, *Result.sVec2#Type)
        Protected.d length = Vec2#Type#_LengthSq(*Vec)  ; << here you calculate the squared length
        If length = 0
          ProcedureReturn #False
        EndIf
        *Result\x = *Vec\x / length  ; << should be Sqr(length)
        *Result\y = *Vec\y / length  ; << should be Sqr(length)
      EndProcedure

• It seems to be inconsistent to define methods for doubles and floats, but returning always a double in functions like Length, Dot, Distance, ... .
  They should return floats as well, when float-vectors are used, to avoid unnecessary type conversions.

[akamicah]

I just looked over it quickly, but I want to mention some issues:

• You define a function RSqr(Number.d), the reciprocal square root. However, the definition is wrong:

  Code: Select all

    Procedure.d RSqr(Number.d)
      ProcedureReturn 1.0 / Number ; << should be 1.0 / Sqr(Number)
    EndProcedure  

• Similar issue in vector2.pbi, here your normalize procedure misses the square root as well:

  Code: Select all

      Procedure Vec2#Type#_Normalise(*Vec.sVec2#Type, *Result.sVec2#Type)
        Protected.d length = Vec2#Type#_LengthSq(*Vec)  ; << here you calculate the squared length
        If length = 0
          ProcedureReturn #False
        EndIf
        *Result\x = *Vec\x / length  ; << should be Sqr(length)
        *Result\y = *Vec\y / length  ; << should be Sqr(length)
      EndProcedure

• It seems to be inconsistent to define methods for doubles and floats, but returning always a double in functions like Length, Dot, Distance, ... .
  They should return floats as well, when float-vectors are used, to avoid unnecessary type conversions.

Thanks! Will push up changes momentarily. There's probably a few other little things I missed as well rushing to put something initial together, so will fine-tooth comb it during the weekend

[Axolotl]

Hi akamicah,
I really wish you much success with your project. But, without demotivating you, I would like to point out that there are already several similar offers on github. What they all have in common is that after an initial phase of euphoria, the expansion and updating very quickly came to a standstill. Please do not misunderstand, it is not a criticism but only a statement on my part.
I have also played with the idea of joining the various github repositories. However, github is absolutely not my thing....
So i will not (be able to) help.
Good luck.

[akamicah]

Hi akamicah,
I really wish you much success with your project. But, without demotivating you, I would like to point out that there are already several similar offers on github. What they all have in common is that after an initial phase of euphoria, the expansion and updating very quickly came to a standstill. Please do not misunderstand, it is not a criticism but only a statement on my part.
I have also played with the idea of joining the various github repositories. However, github is absolutely not my thing....
So i will not (be able to) help.
Good luck.

No worries! The main motivator for this repo is that I would benefit from it so I shall continually contribute to it over time which hopefully avoids the trap others may have fallen in when starting something similar. A lot of code does exist in my own private git server which I do aim to tidy up, and if it's adapted from someone else's code then find the original code on the forums to correctly attribute the works before adding it to the silo library.

As for git - it's not at all a requirement to use git to contribute, and downloading a repository as a zip is always an option for those who don't want to touch git - the purpose of git is to manage the repository, keeping it in existence, and providing an advantage to people who do use git who'd find it easier to add it as a submodule to their own projects

[skywalk]

I applaud your efforts. But, it would be more desirable to have a PB sponsored set of user libraries.
Each having a set of standard tests to avoid simple bug proliferation.
As these libs become more rugged, they could be added to the PB installer.

[akamicah]

[*]It seems to be inconsistent to define methods for doubles and floats, but returning always a double in functions like Length, Dot, Distance, ... .
They should return floats as well, when float-vectors are used, to avoid unnecessary type conversions.
[/list]

Merged an update breaking those methods out of the implementation macro - each will return float now unless a vector of double is being used

I applaud your efforts. But, it would be more desirable to have a PB sponsored set of user libraries.
Each having a set of standard tests to avoid simple bug proliferation.
As these libs become more rugged, they could be added to the PB installer.

As nice as that would be, Fred and co I imagine are in the same boat of time being finite, and that would potentially also remove the open-source aspect which would be a shame

[skywalk]

I'm suggesting something similar to the now open source IDE?

[akamicah]

I'm suggesting something similar to the now open source IDE?

Ahhh gotcha. That'd be really good

[Thorsten1867]

I also have a small silo of PureBasic modules.

https://github.com/Hoeppner1867/PureBasic

[akamicah]

I also have a small silo of PureBasic modules.

https://github.com/Hoeppner1867/PureBasic

Very nice