Update April 10: This is substantially rewritten from the original, to fix a mistake Stephen noted in the comments. I tried just doing strike-throughs of the erroneous or irrelevant parts, but that just made it more confusing.
On May 20, San Francisco will see a partial solar eclipse. (It’s an annular eclipse north of San Francisco, but just partial here.) Maximum eclipse is at 6:32 pm, when the sun will be at 18.9° altitude, 281.4° azimuth; the moon at 18.9° altitude, 281.5° azimuth (figures from the USNO Altitude/Azimuth calculator).
Obviously, the ideal spot for observing and photographing would allow framing the eclipse and Sutro Tower together. Back in January, Burrito Justice used the Photographer’s Ephemeris app to get exact figures for the Mission: 18.8° altitude, 281° azimuth at 6:32 pm PDT:
Based on the USNO calculator, Monday, April 23 should be a perfect night for a dry run. The new moon will set half a degree north of its eclipse track.
On April 23 at 8:36 pm PDT, the moon will be at an elevation of 18.9°, azimuth 282°, or a little over one full-moon width north of the maximum-eclipse position on May 20. Assuming the fog cooperates, it should be easily visible: 7% illuminated, setting an hour and 15 minutes after sunset. At about 9:57 pm, the moon will set behind Twin Peaks at azimuth 293°, still half a degree north of its May 20 track.
The eclipse will be over by sunset/moonset on May 20, though. For a dry-run of framing the maximum eclipse, and then its descent toward the horizon, the best time will be from around 8:30 to 9:00 pm on April 23.
(Top images from Stargazer95050‘s Flickr stream, bottom image from Burrito Justice.)