1) Analytical Hessians for DFT methods would be most useful as it would greatly speed up transition state geometry searches and vibrational analyses.
2) New DFT functionals like, for example, the M06 family of functionals would indeed be very good to have. Functionals which include ab initio correlation like B2PLYP would be useful for use in TDDFT.
3) Better convergence programs can be of big help when dealing with lengthy and/or difficult geometry optimizations. Personally, I think that the addition of the GEDIIS method by Li and Frisch, which performs better than GDIIS and RFO, could make a useful addition.
4) In addition, I would like to request the possibility to calculate CD spectra using TDDFT, since Firefly already contains a very good TDDFT program.
Thank you for your continuing efforts to improve an already great QC program.
[ This message was edited on Sat Apr 16 '11 at 4:21pm by the author ]