[19.0][MIG] survey_next_survey_update_partner: Migration to version 19.0 - #254
[19.0][MIG] survey_next_survey_update_partner: Migration to version 19.0#254adasatorres-tecnativa wants to merge 7 commits into
Conversation
|
/ocabot migration survey_next_survey_update_partner |
|
@adasatorres-tecnativa can you rebase please |
Currently translated at 100.0% (17 of 17 strings) Translation: survey-17.0/survey-17.0-survey_next_survey_update_partner Translate-URL: https://translation.odoo-community.org/projects/survey-17-0/survey-17-0-survey_next_survey_update_partner/it/
d2ccd23 to
2f2a086
Compare
|
@eduezerouali-tecnativa ready |
cristina-hidalgo-tecnativa
left a comment
There was a problem hiding this comment.
Tested locally and code-reviewed.
Migration looks fine and the test suite is green, but the module doesn't work end-to-end on 19.0 — and it's a survey_answer_generation bug, not this PR's.
Chain two surveys, map a single/multiple choice question between them (Country, Tags…), and answer them anonymously (private window / no login — otherwise the answer already has a partner_id and no contact is generated). Submitting the second survey crashes:
_compute_diff_with_origin doesn't str()-coerce the values in the else branch, so a linked choice line with no suggested_answer_id yet (prefilled question, not answered while navigating between pages) gives current_value = False and get_diff() blows up on False.splitlines(). Since diff_with_origin is a batched stored compute, that one line kills the recompute for all of them → they stay False → _mark_done never propagates any changed prefilled answer to the partner (only brand-new questions get through).
IMO, it should be fixed first (or in the same series). Worth extending the test here to modify a prefilled choice answer too — the current tour only covers forward char_box edits.
2f2a086 to
6ce3a43
Compare
6ce3a43 to
ae9308a
Compare
@Tecnativa TT64015
@pilarvargas-tecnativa @eduezerouali-tecnativa