Saturday, July 29, 2017¶
I deployed my changes for #1982 to their production site.
And again, as already Friday, June 23, 2017 and Wednesday, July 26, 2017, we want to re-import master data from TIM without loosing accounting data. This time it is mainly because we want to have the legacy PAR->IdPar numbers as the ref of each Course.
Here is the general approach:
make snapshot a (before upgrade)
python manage.py run a/restore.py and repair manual migration issues
make snapshot b (after restoring a into new version)
Run ./init_demo.sh in order to (correctly) re-import all data from TIM. This will destroy all bookings which were already done.
make snapshot c (with correct partners but without bookings)
copy the following files from b to c (overriding files in c):
finan_*.py sales_*.py ledger_*.py accounts_*.py products_*.py vat_*.py ana_*.py
python manage.py run c/restore.py to put it all together.
I wrote a run script for manually converting some of their voucher dates.
I fixed some exceptions which occured when
Movement.get_mti_leaf() returned None. This situation was
exceptional, it came because the VAT declaration was lost during data
migration, but its movements were there.
TODO: write a data checker for Movement which detects this and removes the movements.
Lino too big for ReadTheDocs?¶
I guess that if I file an issue, the readthedocs team will tell us to split the book into smaller parts. But I don’t see how we could do that. The two biggest beasts of the book (Developer Reference and API) are recursively interdependent: module docstrings usually refer to the specs page which describes them, and specs usually refer to elements of the API. And note that the Lino specs are far from being complete. And note that we plan to include Voga and Welfare into the book as well. Hmm…