On 2011-03-17, at 15:16 , chevalma@iro.umontreal.ca wrote:

I'm working on a little sketch of my predoc report and would like some feedback. I'm just trying to establish an outline of what I will be discussing. As such, I wrote a table of contents and I've begun to gather a bibliography of papers that I will probably mention in my related work section.

Could you take a look and make suggestions to improve it? Do you think that there are things missing from the outline, or that I should organize things differently? Can you think of other papers I should be citing?

As a reader, what I'm looking for in any predoc report is the following :

1) What is the problem? 2) Why is it important / why should I care? 3) Why hasn't it been solved before? 4) What do you propose to do about it?

Then, the committee's job is to make sure that #1-3 are understood by the student, and that #4 is realistic for the amount of time a PhD normally takes (often, you could get a few PhDs out of a single predoc report:)).

Section 1 should answer both #1 and #2. From your plan, I'm not entirely sure how it will achieve that though. Maybe you can clarify your thoughts a bit. Also, make sure #1 is crystal clear. It's often overlooked, IMHO. Obviously, section 2 answers point #3. Just make sure that for each paper (or group of related papers) you point out why it doesn't really solve your particular problem. In section 3 answers point #4, but to me it looks a little bit fuzzy at the moment. I'm not sure why metacircularity shows up as a subsection, since AFAICT this is Erick thesis. Maybe it would help if you could just summarize your thesis in one sentence. From our earlier discussions, it seems that your focus will be on novel optimistic optimizations for javascript (or dynamic languages in a broader sense, but that will have to be determined) for performance. So your thesis is something like 'there are opportunities to exploit dynamic information in new ways that would reduce the running time of JS programs', but I find that too genera. I'm sure you can come up with a better/more precise one. Also, I would expect a little more discussion about that particular aspect of your work (the new optimizations) than about Tachyon itself, but from your outline I can't be certain this will be the case.

Finally, what's 4.3 supposed to be about? Personally, I can't see how this relates to your thesis, but my opinion is based on the title alone...

I think that perhaps tracing JITs deserve a section just for them under related work, seeing the large amount of publications about them.

For me, the particular structure tends to emerge as I write the related work section. As long as it's clear and coherent, it's normally fine.

Bruno