This is pretty much a dump of what I’ve been working on during making this port. It also contains some dev screenshots.

December 1, 2016

Day 53 + image of the day

fixed a problem with texture that had id = 0 not being applied :-) it was a font texture (used at least for in-game text), so some text was not visible due to this reason

Image of the day:

November 29, 2016

Day 51

was changing long to int32_t in ABL related files… well externs are evil! consider:

1.cpp

int myVar;

2.cpp

extern long myVar;

good luck looking for a bug…

ABL uses lots of them!!!

even if you think you’ve correclty changed all longs to ints, there is such code as:

long my_stack_var;
memcpy(myVar, my_stack_var, sizeof(long)); // well this you may catch by searching for "long"
// but consider this
memcpy(myVar, my_stack_var, sizeof(my_stack_var));// :O

I hope I did not leave such cases….

Also was changing(long -> int) in all structures that are saved in order to surpport saves from 32bit systems

Finally, fixed nasty bug when repair truck sometimes was not appearing. As I see, it is uninitialized variable isOnGui which prevented it from being drawn

when it happened that it was false, repair truck tried to be drawn as alpha transparent oject, but in fact was not drawn because there was no branch that was drawing objects with just MC2_DRAWALPHA flag set. (see comments in commit for more explanation)

Look like Windows version always initializes memory with 0xff an so they did not have such a problem. Should I do the same? :-) just to repeat all their issues :-)

no I’ll expose all of them and fix them… maybe :-)

Ok, will have a cup of tea and go to sleep, after all I have a birthday tomorrow! (that is because I am writing it on 28 though this blog records is for day earlier)

November 29, 2016

Day 52

fixed memory leak in gosAudio_CreateResource

in training mission 5 "terminationResult" variable was screwed (see mission.cpp 667 - 676 lines)

apparently guys had a strange sense of humor and added playing of win/lose tunes if terminationResult was not as one might expect to find out ABL issues, and apparently training mission 5 has them.

I do not know if it had them before, or they we caused by my recent changes in abl code (long->int), I hope not. Will have to check tough

November 26, 2016

Day 50

finished initial draft sound implementation, there are some wav files which cannot be loaded neither by SDL nor by SDL_mixer, also noise sounds that repacked have to be extracted.

made couple of small fixes like avoiding triggering assert() when dividing by a very small number, happened because currently frame time is faked by me as 16msec :-), and others

even finished E3 mission :-) from .. maybe 10th try, because of anoying asserts and PAUSE()/STOP() triggerings. some strange issue still remains with trying to render text which is NULL

November 25, 2016

Day 49

proceeded with sound, almost can play game with all sounds available, though still lack a lot of sound support stuff and resource loading could be smarter.

November 23, 2016

Day 48

started to add sound support. looks like signed to unsinged conversion fails to work but mono/stereo and frequency conversion works, hmmm

no screenshots in near future, I expect :-)

November 19, 2016

Day 47

added conversion between rgba -> bgra -> rgba when locking and unlocking, because CPU code expects textures in BGRA, so I have correct colors now

also fixed very interesting issue.

fors code was like this:

unsigned long s_textureHandle[] = {-1, -1, -1, -1};

clang was complaining that there is conversion or something like this, so I changed this to:

unsigned long s_textureHandle[] = {-1u, -1u, -1u, -1u};

and then today I’ve got a crash because of this conditional:

if(s_textureHandle[i] == -1) {
}

It was giving false, what happened in this case is:

-1u got converted into 0xffffffff, and promoted to long.
inside condition expression: -1 got promoted to unsigned long with sign extension, so first: it became -1 long, and then converted to unsigned 0xffffffffffffffff, and suddenly it all broke

before it worked, of course. because in construction of s_textureHandle[] -1 got converted to 0xffffffffffffffff as well.

phew, I have to be more carefull with those longs

November 19, 2016

Day 46

trying to figure out why some textures do not have nice blue background…. :-/ like ones in solo mission on single player option map.

but in-game minimap has this nice background and windows version also has it

November 14, 2016

Day 45 + image of the day

implemented proper (well.. I mean, it works sometimes) texutre lock/unlock

figured out why mouse cursor is not drawn sometimes (mc2 mouse is processed in a separate thread. Something like this variable was set or not set depending on some strange reasons)

we shoulld always assume it is not in a separate thread and that is it!

spent whole evening tracking strange issue when mechs would not go where I told them… but would do it if I said to shoot, apparently I had 2 errors in timeGetTime()

and overall I was wrong by 1e3 in 2 places :-), thanks gdb and watchpoints, now camera also works better and I even finished a mission, though, it crashed on a mission end screen :-)

so, quite a productive evening today. thank you all :P bye!

Image of the day:

November 12, 2016

Day 44 + image of the day

continue with various fixes.

one that was quite scary is memery block filling with NANs. when first memory was cast to long* but number of iterations was divided only by 4, ..who cares about 64bit long in 2001

finally I can play a game now! though mech would not move where I tell them, only start moving and then stop again. It is interesting, that they move if they need to come closer to shoot (when I aim cursor at the target and click)

Image of the day: