rewrite arrays to structs

[turican0]

I'm glad that the project continues and sometimes I help with it (turican0). Now I'm rewriting the array into structures, this is necessary for further orientation in the code, but it will cause a number of regressions. When it's all done, I'll start next bug hunting and repair code.

[GrimSqueaker]

Hi Tomas,

thanks for your efforts. It is great that you can continue. I would like to make it compile and work on 64bit in the future and the structures are a prerequisite for it. In this area you are the expert. For all of us this is a hobby, so there is no pressure and only fun. :-)

Regarding bug hunting: I was a bit too fast to merge your changes and they introduced a regression, see Tim (@thobbsinteractive) and my conversation here: https://github.com/thobbsinteractive/magic-carpet-2-community-patch/pull/15 To summarize: The fireball sprites are broken when hitting buildings. One of the following commits introduced it:

$ git log 2b098da3^..aac4f686
commit aac4f68699453da60c014428242bf01be87fd7e3
Author: turican0 <41465132+turican0@users.noreply.github.com>
Date:   Tue Jan 26 20:04:27 2021 +0100

    rewrite nest arrays to sctruct

commit 3533ac298561a0fd078c2bbb9e1473666fc0d4db
Author: Tomas Vesely <turican0@gmail.com>
Date:   Tue Jan 26 15:27:53 2021 +0100

    next fixes

commit 3de3e51912699518b86acaeec5034e82d6f02669
Author: turican0 <41465132+turican0@users.noreply.github.com>
Date:   Mon Jan 25 18:42:54 2021 +0100

    modify info text

commit 2b098da3fb4b9294456955bcd629368826458451
Author: turican0 <41465132+turican0@users.noreply.github.com>
Date:   Mon Jan 25 18:37:32 2021 +0100

    add informations for next testing

Cheers and thank you, Sebastian

[turican0]

Hi, thank you for report. I hope yours cooperation on this project. Now I want maximalize rewrite code for better readibility. This phase will be long 1-4 week. During this time, I would only fix bugs that prevented the engine from running. So that my rewriting does not hinder things aka rendering changing, test run on other OS, etc. I will find errors until after this phase, because it is now faster to look for all errors at once, not each one separately.

I would divide bug hunting on small and big.

Small bug hunting - fix first level, not visible problems - time for this cca 1day(plus find bug fixes) Big bug hunting - replay all game, every level begin(compare data with dosbox version), test all spells in game(compare data with dosbox version), test all options in game, etc. - time for this cca 30day(plus find bug fixes)

After rewrite structures I make only small testing. I would like to agree on a Big bug hunting with you. It is advisable to do this only after maximum code cleanup and rewriting.

(I apologize for my English, I don't speak much :)

Tom.

[GrimSqueaker]

Hi Tom,

yes, rewriting for better readability is a really good thing to do. Until now I have just limited my changes, because I do not know if you for example still need the memory address references in the function and variable names. I guess that this makes it easier for you to compare to the original in dosbox, does it?

I am not sure, if I can help you with the bug hunting, because some areas of the code are still disabled on Linux (they do not compile because of type problems) and I do not know how to compare to the dosbox version. I'll try to get the Linux build as far as possible so that I can support you as good as I can. Please let me know how I can help you.

Cheers, Sebastian

[turican0]

Hi Sebastian,

Yes at now is easer have memory references in name(or in comment after name?) of functions for compare with dosbox.

For bug hunting I use https://github.com/turican0/dosbox-x-remc2. I modify original DOSBOX-X to extract memory parts at some conditions. This memory parts I compare with memory parts in remc2 at same conditions and find position of code creating differences(bugs). After rewriting some structures I will create little manual for this. If you find non-compilable code in linux parts of code, write it please in new issue. Maybe some will be fixed during rewriting.

By, Tom

[GrimSqueaker]

Hi Tom,

thanks for the explanation. It would be great to have a small manual for the memory comparison.

I think I will deal with all the Linux related compile problems step by step. At the moment I do it like this: Play a level and check the printout (the "FIXME: types @ function"). Fix the types and play again to check that I did not break anything. When starting with the Linux build I did decide to comment it out only for Linux in order to quickly have a working build. But the "play a level" part take a while. :-) But I do appreciate if you fix some of them during your rewrites. The GitHub continuous integration will tell you if a pull request breaks the Linux build. That is why I did implement it as a first thing. This way you can work on either system and easily find out if it still works on the other.

Cheers, Sebastian

[turican0]

Yes, after I rewrite and fix some bugs, I find parts witch FIXME: types, and try fix this code. Manual to compare with dosbox-x I write, when I do this action.

By, Tom.

[turican0]

Can you give me the rights to add labels and close issue? (I will not manipulate with git code) Thank you, Tom.

[GrimSqueaker]

I have added you as a collaborator. I think you got a invitation for this.

[turican0]

Thank you, but unfortunately I still can't assign labels :( Tom.

[turican0]

It already works, I had to confirm the note, I was not on the email in the morning :)