Holography records not a 3D image but an encrypted wavefront. To determine what object must have caused that wavefront, we must solve the inverse problem. Most of the time, inverse problems are singular, but in certain very simple cases they are not and an optical computer does that very well. In others, it is disastrous. So, optical computing has been part of holography from its beginning in 1947. It was not until 1962 with VanderLugt's work on Fourier Optical pattern recognitiion that holography became part of optical computing. In this paper, I review both directions of that continuing intercourse between hologrpahy and optical computing. Neither requires the other, and both are sometimes best without the other. But there remain a few cases where the fit seems ideal. The processing and recording and modification of spectral-temporal patterns is the test case I will use to illustrate their constructive relationships. Here are some examples: Spectral recognition is best done in the optical domain by holography. Temporal pulse shaping is best done in the optical domain by holography. Temporal pulses are best recorded by holography.