Closed ribbanya closed 7 months ago
ar extract
, elf disasm
... and that's about it. There's no split step. The build system is just calling mwcc and comparing with objdiff.dtk dol diff
over to objdiff proper.gx
, gxD
, etc.) and switch to "profiles" later.To expand on point 1 and 2, we're dealing with two pretty different paradigms:
It would be possible to consider what they have in common and extract shared code, but I'm leaning towards the idea that it's better to keep them as two separate more focused project structures that happen to duplicate some code / ideas.
I'm happy to hear arguments otherwise, though, especially if there's ideas on how we could support both without complicating project.py
even more.
That makes sense. So for now do you think the build system should live entirely on doldecomp/dolsdk2001 without trying to abstract it?
Yeah, for now. And I think dolsdk2001 should aggressively trim out everything that's unneeded.
Okay, cool. Closing this as it's no longer related to the dtk template.
So I've been messing around with this branch for a couple days, trying to support dolsdk2001. I've gotten it to extract the archives, build the source files, and compare them using
objdiff-cli
. Profiles (release vs. debug) aren't fully supported yet.Proof of concept
Branch changes
Inquiry
At this point, I'd like to stop hacking this together and instead go about it in a way that can be upstreamed.
/config
and circumvents the split/symbols features of dtk. Should dtk (rather than just the template) be modified to support libraries?OK
check, or should that still be handled byobjdiff-cli
?-sym on
?project.py
and dtk make a lot of assumptions about the existence of a linker.