My initial goal is to document the CTF format and forge this knowledge into an useful and sober API. Therefore, I contacted people at dtrace.org, one of the very few groups of people that might be interested in my work. I briefly introduced myself and my project, stating my objectives and asked for commentary and opinions.
Only a few moments after my mail, I got a reply from Robert Mustacchi. To summarise his input:
- to take a look at D-Trace print() action and mdb’s ::print dcmd
- he suggests to take the top-down view of the problem: try to prototype the CTF tools like ctfmerge, ctfdump or ctfconvert and see what possible API they might need
- he finds the reading/lookup API to be good enough in the illumos libctf
The first point seems like a reasonable thing to do (along with the LLDB default settings), but it has to wait until we have working libctf. I like the top-down approach idea and will suggest it to George (my mentor).