Wow! It has been quite a while here in Taiwan. It looks like I am nearing the end of my internship. In fact, I have only two days left (today, and tomorrow) before my internship comes to an end.

Okay first of all, recently, I got a pretty good opportunity to join the HitCON conference. It just so happens that this company ITRI is one of the sponsors for the event. This will be my first conference that I am able to attend, so I am pretty excited. I am still a bit nervous about what I would be able to do there, but I think I'll wait and see. I don't think this opportunity really happens often, so I anticipate for this event.

I guess I don't have much else to say, so I guess I will summarize the trip. From the very beginning of this trip I recall that one of the main issues I'm having was my language barrier. Gradually, that has become less and less of an issue. Now, even though I still feel like a foreigner sometimes (that might be a different story), I definitely feel that if I had to only communicate in Chinese, I would be fine (that technically was what happened the last nine weeks). Honestly, I think the internship was a bit slow to start, and probably I didn't finish much on the trip. However, I did definitely gain an experience that I could not have otherwise ^_^. It is definitely a fun experience, and I am both relieved and saddened that I will be leaving shortly. Relating to AFL fuzzer, I think it currently doesn't seem to fully work (at least the stuff that I've been working on). I dunno, I've had some pretty cool ideas, but is it perhaps that these won't work in practicallity? The author of AFL did say this:

The state correlation approach sounded very sexy on paper, but ultimately, [it] made the fuzzer complicated, brittle, and cumbersome to use; every other target program would require a tweak or two.

So I guess it has already been tried. Maybe perhaps there might be a light to the idea, but perhaps very difficult. Now looking in light of this, it maybe that the person who picks this might be entering into a deathtrap. Perhaps too much resources are being spent to solve this problem? Is there even more optimal solutions to AFL? After spending much time on AFL I realized that AFL is already very optimized and pampered piece of software. Perhaps trying to improve on it is just moot point?

Hehe. I've started out just thinking about writing a summary, and I go on a pretty long rant. Oh well. I guess that's about it.