Table Out1 has translated versions of fields AA..AZ from table In1.
Table Out2 has translated versions of fields BA..BZ from table In1.
table Out3 has translated versions of fields X..Z from table In2, and the translation depends on some of the AA..BZ fields, in both translated and untranslated forms.
We'd end up with views that depended on other views and after expanding had like a dozen self-joins to add more fields that weren't included in the base view. And views that were selecting a big list of columns unchanged, plus adding a couple.
But somehow despite being an absolute ^%#@!^&$! it still worked better than the pointy-clicky ETL tooling we'd been using previously. Mostly because of the run->debug->modify cycle time being faster, and only a little because of avoiding pointy-clicky stuff.
Table Out1 has translated versions of fields AA..AZ from table In1.
Table Out2 has translated versions of fields BA..BZ from table In1.
table Out3 has translated versions of fields X..Z from table In2, and the translation depends on some of the AA..BZ fields, in both translated and untranslated forms.
We'd end up with views that depended on other views and after expanding had like a dozen self-joins to add more fields that weren't included in the base view. And views that were selecting a big list of columns unchanged, plus adding a couple.
But somehow despite being an absolute ^%#@!^&$! it still worked better than the pointy-clicky ETL tooling we'd been using previously. Mostly because of the run->debug->modify cycle time being faster, and only a little because of avoiding pointy-clicky stuff.