I had a paper accepted at a conference in theoretical computer science. However while revising the final version I found an error in the proof of an important lemma. Although I believe the error is fixable, It will require some effort. I will not be able to fix the error before the deadline for the final version.
1) Should I withdraw the paper?
2) Should I leave the lemma without proof, and add the corrected proof at the final version.
What would be the best thing to do? Would I be able to resubmit the paper to some other conference? What would be the damage for my reputation?