| 204 | |
| 205 | ==== Observed file formats and sample data: ==== |
| 206 | |
| 207 | |
| 208 | - Outer join has been observed "in the wild", e.g. the |
| 209 | [http://haiti.resource-finder.appspot.com/export?subject_type=hospital Google hospital data for Brazil] |
| 210 | has one level of join with a 1-1 valence, but adds a second level by using... |
| 211 | |
| 212 | - Lists in cells. |
| 213 | |
| 214 | - !OpenOffice spreadsheet export saves sheets in one spreadsheet to separate files. |
| 215 | OO spreadsheets were used to load Southern California hospital data into the |
| 216 | Humanity Road database, and these had separate files associated via UUID. |
| 217 | (The load was done by a script that did dal calls and constructed its own |
| 218 | org_site records. The secondary files were not linked OO sheets -- they just |
| 219 | had a column for the associated hospital's UUID.) |
| 220 | |
| 221 | - Need example of Excel exported CSV files. |
| 222 | |
| 223 | We should stop here and decide: |
| 224 | |
| 225 | ==== Which of these formats do we expect to encounter and want to support? ==== |
| 226 | |
| 227 | Besides the observed forms, please propose others that we want to support: |
| 228 | |
| 229 | - (Comments please!) |
| 230 | |
| 231 | ==== Which should we implement in the initial version? ==== |
| 232 | |
| 233 | Yes, we should keep an eye on the general case and make this easy to extend. |
| 234 | Consider this to be prototyping (that may happen to have a specific use in mind). |
| 235 | |
| 236 | - (Pat:) Single file, outer join, allow lists in cells. |
| 237 | |
| 238 | - (Comments please!) |